I promised a while ago that I would provide a text version of my talk at DevConf, for people who couldn't make it and because sitting through a video of me standing up there going on and on doesn't really make for good followup discussion.
Because it grew rather long, I think it works best as a web article, which you can find on Fedora Magazine at http://fedoramagazine.org/?p=1236.
Also because of the length, I'm posting it installments. This part is the background, next part is some of the things we are doing, followed by the panel discussion, followed by Q&A. Whew; that's a lot.
I can post the text here if people think it'd be really helpful to do so, (although I'm inclined to think that would be unwieldy), and in any case I will take questions, comments, complaints, in any media including replies here, on the article, on the social media, or at any bar or coffee shop within walking distance of Boston's MBTA.
Hi Matthew,
On Wed, Mar 19, 2014 at 11:53:27AM -0400, Matthew Miller wrote:
I can post the text here if people think it'd be really helpful to do so, (although I'm inclined to think that would be unwieldy), and in any case I will take questions, comments, complaints, in any media including replies here, on the article, on the social media, or at any bar or coffee shop within walking distance of Boston's MBTA.
It would be nice if you could post a link to the next installments on the list, otherwise interested people (like myself) might miss them.
Cheers,
On Wed, Mar 19, 2014 at 05:27:20PM +0100, Suvayu Ali wrote:
I can post the text here if people think it'd be really helpful to do so, (although I'm inclined to think that would be unwieldy), and in any case I will take questions, comments, complaints, in any media including replies here, on the article, on the social media, or at any bar or coffee shop within walking distance of Boston's MBTA.
It would be nice if you could post a link to the next installments on the list, otherwise interested people (like myself) might miss them.
I definitely will. Thanks.
Matthew Miller mattdm@fedoraproject.org writes:
Because it grew rather long, I think it works best as a web article, which you can find on Fedora Magazine at http://fedoramagazine.org/?p=1236.
Because really, even though many people think the base OS is now boring, it’s far from done, and there is a lot of innovation going on at that level as well.
What do you consider as "base OS"?
And I don`t think it`s boring. I think it`s bad that
+ you find more and more stuff running
+ that it`s difficult to find out what this or that process does which wasn`t there before the last update/upgrade
+ that a lot of what is running is something you never need and
+ that it is usually difficult to impossible to get rid of stuff you don`t need.
And on top of that, what is the Fedora-way of replacing gnome --- which I find totally useless --- with fvwm, which perfectly does what I want? It`s only one example, and you can figure out how to do it. But the point with this is that Fedora lacks flexibility. You get what you get and then have to go through a lengthy process of getting rid of stuff and of somehow getting to work what you need, like fvwm.
On a side note: The installer sucks, just try to do one of the most basic and important things with it: Partitioning. And when you managed that, you can`t start with a minimal install and install just what you need.
I was at a large cloud conference a while ago, and almost nobody was using Fedora, and so I asked people why they chose the distribution they are building their stuff on, and why they didn’t choose Fedora. Almost universally, the response wasn’t “What I am using is great!” — it was “Oh, I don’t care. I just picked this, and that’s what I’m using and it’s fine.”
I`m not one of these people. Thinking like that, they don`t need a Linux distribution; they can as well use Windoze or Macos.
Even if there really are a lot of interesting things going on, people who are trying to actually do things with the distribution don’t attach much importance to them.
Not paying attention to the soft- and hardware that is the basis of what you`re actually doing is a recipe for failure, and for making things difficult on yourself.
One thing Fedora shines with (so far) is reliability, and reliability is one of the requirements I have. I have been using Debian for almost twenty years until they messed up badly with their brokenarch. Doing that put Debian out of the question once and for all because they failed that requirement miserably beyond believe.
Please do not make the same mistake with Fedora. Switching to another distribution is a painful process.
I`m saying that Fedora shines with reliability because I have not had a single crash or freeze since I switched to Fedora over a year ago. And I`m saying "so far" because it`s only a bit over a year now since I switched. And twenty years is a long time. I`d be happy to use Fedora for the next twenty years. Will Fedora be up to that kind of challenge?
But I find the reliability I`ve seen so far with Fedora very exciting and I am very thankful for it. It is a great accomplishment of Fedora.
So Fedora works. I had to make a pick when I kicked Debian out, and since someone had told me about Fedora, I looked at the web pages and decided to try it. It has it`s disadvantages and quirks and stuff I really don`t like, but it works.
Fedora is great. I put it on the same hardware I had Debian on and was immediately impressed that things are running a lot faster than they used to. I was really pissed when the upgrade from F17 to F18 failed --- or seemed to fail because there was no indication of any progress during the application of SELinux policy rules, which lead me to press the reset button after an hour or two --- but it was possible to repair the damage. And repairing the damage was less painful than switching to another distribution.
So yes, Fedora is great, and I`m impressed, and I have to come to like it. I know why I`m using it, and you could even say "I just picked this and it`s fine".
But I picked it after good consideration, and the consideration does not stop at some point. And I do care: If it should become not upgradeable or unbearably unreliable at some point, I`ll kick it out just like Debian. I don`t need a recipe for failure, I need my computer to work.
Also: the base OS has developed to the point where it has become uninteresting.
Perhaps not: It is possible that the expectations have changed.
Ten years ago people were probably much more willing to accept that their soft- or hardware doesn`t work and that their computer crashes or freezes every now and then than they are willing to accept this nowadays. Some people still don`t care.
Look at different distributions and you may find that they are all using more or less the same software. It doesn`t matter whether I use distribution X or Z or V, they all come, for example, with emacs, and if not, I`ll just compile it myself.
People probably expect things like that. But fail to be great by making a distribution that crashes/freezes their computer every now and then, and people will not use it.
People use this or that distribution because it`s great. Only their expectations have changed and they don`t notice anymore that what they are using is great. They take it for granted.
And by not paying attention, they create their own recipes for failure. Emacs is great, but when the rest of the system doesn`t work, I cannot do what I actually do.
So let people see how Fedora is great, and remind them. You could consider it a point in which Fedora has (and other distributions have) failed that people think it doesn`t matter which distribution they are using.
In turn, this leads to a shift in the balance between the effort to get software into a distribution and the reward of doing so. It used to be that if you had open source software, and you could convince the distros to get your get your software into a distro, that’s how you knew that you had arrived.
I have always wondered how people manage to create packages, for Debian or Fedora. I looked into it because I would like to provide packages, and I found it requires an insurmountable effort. You start with "I have written this software" and get to "I would like that ppl use it, and to make that easy, I`d like to make a package". Then you try to find out how to do that and that`s where it ends: It`s just too difficult.
Instead, you put your software on github.
I’m really big into the idea of incremental improvement. As long as we have a plan and are going forward and making some difference each release, that’s good. I know this sometimes makes people working on new ideas frustrated — for example, removing sendmail from the default install. I’m happy to get sendmail out of there, and I think it doesn’t belong for a lot of cases, but if it takes us two or three releases to remove it, eh, it’s okay.
That decision came across as "removing an MTA from the default install". I don`t know if you`re saying that something is now replacing sendmail or that there is no MTA when you do a default install.
If it`s the latter, no MTA at all, then it was a very bad decision. A system without MTA is not functional.
So that’s part of what Fedora.next is: to look at our mission and decide what more we need to do to make it happen.
Let me simply ask what "community" is supposed to mean in the mission statement (which I can`t quote here because it`s an image).
I am asking this because nowadays, everything is a "community", to the point where that word doesn`t mean anything anymore. What, who and where is this "community" in this case, who are the members of it and how does one become a member?
The mission statement seems to say that "the Fedora Project" itself is a "community", one that is "collaborative". Isn`t being "collaborative" implied in being a "community"? I wouldn`t say that all "communities" are necessarily "collaborative" --- yet in this case being "collaborative" looks like it could be a very substantial factor, constituting "community" in the first place, bringing some redundancy about the mission statement.
Without redundancy, the statement may boil down to "let`s advance FOSS". The point of Fedora.next would then be to figure out how to do that "better" than before.
What does that have to do with a Linux distribution called Fedora? The next step might be to cease making a Linux distribution so that the freed resources may be dedicated to other activities which more efficiently, i. e. "better", "advance FOSS".
Like you said, nobody cares anymore what distribution they are using, and it`s a topic that has become /really/ boring --- so why not do something exciting that more efficiently advances FOSS than making one?
Most everything we are talking about in Fedora.next is additive. We’re not saying that you can’t do certain things in Fedora anymore, but we are saying that we may need to add some things to make the goals of the mission come true.
I seriously don`t like this statement. I don`t like it not because it brings about the very idea it seems to be denying but because I don`t like the idea it brings about.
That`s a political strategy, telling people that "we" aren`t going to do what "we" are going to do because "we" are going to do something else. What actually goes on then remains unnoticed, and not long after every time after having seen or heard this statement, you look back and see that what the "we" said they wouldn`t do is what they actually did.
I don`t buy that anymore.
Of course, we also need to look at how we can make what we are doing well even better, possibly by changing some of our focus and certainly by improving our processes.
Listen to the users, if you want to make a better distribution.
No one has really been able to successfully make software which goes on top of Fedora and keeps up. Can we make that easier in some way?
Well, I have. It`s running here right now, since F17. I`d make a package if it was easy to make one ...
Maybe we could do a better job by letting people put their packages in
Assuming that they can make packages ...
to some level in an easier way, and then after they’re included in the project look at improving them. (Update This is actually in progress in the Enviroment and Stacks Working Group now, in the “Fedora Playground” repository. Cool!) And maybe also, we could focus more on packages where it really matters that they’re packaged well, and decide that with others, well, eh, that’s probably going to be crap forever and we’ll do what we can to make it as useful as possible to people.
All packages need to be packaged well. I don`t mind having additional repos with crappy packages that need to be explicitly enabled.
But making crappy packages the standard obsoletes having packages (beyond a minimal set that gets you started) because you`re better off pulling the sources and compiling them yourself, rather than installing a crappy package with which you have no way of knowing what it might mess up.
Hm. After reading the article, I still don`t really understand what Fedora.next is about. It seems to try to somehow put together some questions like "How can FOSS be advanced more efficiently?", "What can be done to make Fedora a better distribution for everyone?", "Do we want more people to use Fedora?", "How do we get more software into Fedora?", "How do we get more people to contribute?", "How can we decide what we want?" and "How can we struggle less to get what we want?".
I think I`d like to see an article with questions like this which lays out what answers to these questions are currently in place. There probably aren`t very many people who know the current answers. Such an article would also need to explain for each question why it is necessary to ask it now.
On Fri, Mar 21, 2014 at 04:19:05PM +0100, lee wrote:
Because really, even though many people think the base OS is now boring, it’s far from done, and there is a lot of innovation going on at that level as well.
What do you consider as "base OS"?
It's somewhat nebulous, but, as a general working definition, the system stuff below the applications layer. (Not in the OSI sense.)
And on top of that, what is the Fedora-way of replacing gnome --- which I find totally useless --- with fvwm, which perfectly does what I want?
It sounds like you want to do a minimal install and then add up from that. I think you will benefit from this effort in that the minimal install will be better defined and curated.
It`s only one example, and you can figure out how to do it. But the point with this is that Fedora lacks flexibility. You get what you get and then have to go through a lengthy process of getting rid of stuff and of somehow getting to work what you need, like fvwm.
Sure, I would agree that this isn't a strong suit, particularly with the all-or-nothing way RPM dependencies currently work. But on the other hand, I'm not sold on it being a huge problem. If you know what you are doing, it's not that big of a burden, and I'm pretty sure that the intersection of people who want this and people for whom it is easy with Fedora as it stands is quite large.
On a side note: The installer sucks, just try to do one of the most basic and important things with it: Partitioning.
Saying something "sucks" isn't very helpful. Not only is it needlessly negative, it is intangible. Name a real problem and we can talk about it.
On the specific you do give, I'm pretty confident in saying that you're actually wrong. Storage is hard, and the new anaconda contains the most sophisticated and powerful GUI partitioning tool ever made. It's just a really difficult area to get perfect. Expect this to continue to improve with each Fedora release as the design is refined and bugs are shaken out.
And when you managed that, you can`t start with a minimal install and install just what you need.
You can start with a pretty basic install in the GUI. If you need more than that, you really should be looking at kickstart, and maybe even producing your own spin.
I was at a large cloud conference a while ago, and almost nobody was using Fedora, and so I asked people why they chose the distribution they are building their stuff on, and why they didn’t choose Fedora. Almost universally, the response wasn’t “What I am using is great!” — it was “Oh, I don’t care. I just picked this, and that’s what I’m using and it’s fine.”
I`m not one of these people. Thinking like that, they don`t need a Linux distribution; they can as well use Windoze or Macos.
Yet these people were absolutely running Linux. Just not ours.
One thing Fedora shines with (so far) is reliability, and reliability is one of the requirements I have. I have been using Debian for almost twenty years until they messed up badly with their brokenarch. Doing that put Debian out of the question once and for all because they failed that requirement miserably beyond believe. Please do not make the same mistake with Fedora. Switching to another distribution is a painful process.
I don't know what that specific problem was with Debian, but we are certainly working at increasing reliability, particularly through automated testing.
But I find the reliability I`ve seen so far with Fedora very exciting and I am very thankful for it. It is a great accomplishment of Fedora.
As I started with: Fedora is awesome.:)
Also: the base OS has developed to the point where it has become uninteresting.
Perhaps not: It is possible that the expectations have changed. Ten years ago people were probably much more willing to accept that their soft- or hardware doesn`t work and that their computer crashes or freezes every now and then than they are willing to accept this nowadays. Some people still don`t care.
Actually, the change is the opposite, at least in the context I am talking about. Check this out: https://www.usenix.org/conference/lisa13/how-netflix-embraces-failure-improv...
[snip]
I have always wondered how people manage to create packages, for Debian or Fedora. I looked into it because I would like to provide packages, and I found it requires an insurmountable effort. You start with "I have written this software" and get to "I would like that ppl use it, and to make that easy, I`d like to make a package". Then you try to find out how to do that and that`s where it ends: It`s just too difficult. Instead, you put your software on github.
Absolutely. We need to make that easier.
That decision came across as "removing an MTA from the default install". I don`t know if you`re saying that something is now replacing sendmail or that there is no MTA when you do a default install. If it`s the latter, no MTA at all, then it was a very bad decision. A system without MTA is not functional.
Sure it is. Again, hyperbole doesn't really help. In fact, in many situations, an MTA doesn't do you any good, as your network won't allow it to do anything useful remotely (true now at most big companies and on most big ISPs) and local delivery goes into the black hole of root's mailbox unless you configure it. And if you're going to configure one, installing one isn't a significant extra step. Plus, since there are multiple different MTAs, this is back to your choice of fvwm vs. gnome -- if we have a default you don't want, you're going have to remove it to put in your choice.
So that’s part of what Fedora.next is: to look at our mission and decide what more we need to do to make it happen.
Let me simply ask what "community" is supposed to mean in the mission statement (which I can`t quote here because it`s an image).
https://fedoraproject.org/wiki/Overview#Our_Mission
"The Fedora Project's mission is to lead the advancement of free and open source software and content as a collaborative community."
I am asking this because nowadays, everything is a "community", to the point where that word doesn`t mean anything anymore. What, who and where is this "community" in this case, who are the members of it and how does one become a member?
"Collaborative community". That means it is all the people who work together on the project. You can become a member by saying you are and doing something. That's all there is to it. (Although, practically speaking, you will also want a Fedora account if you don't have one.) See https://fedoraproject.org/wiki/Join for a lot more starting material.
Listen to the users, if you want to make a better distribution.
Sure. Continue speaking up -- we are listening.
No one has really been able to successfully make software which goes on top of Fedora and keeps up. Can we make that easier in some way?
Well, I have. It`s running here right now, since F17. I`d make a package if it was easy to make one ...
Maybe we could do a better job by letting people put their packages in
Assuming that they can make packages ...
Excellent point. Not requiring packages is the next level.
Hm. After reading the article, I still don`t really understand what Fedora.next is about. It seems to try to somehow put together some
Stay tuned for the next section -- it might clear some of these things up.
questions like "How can FOSS be advanced more efficiently?", "What can be done to make Fedora a better distribution for everyone?", "Do we want more people to use Fedora?", "How do we get more software into Fedora?", "How do we get more people to contribute?", "How can we decide what we want?" and "How can we struggle less to get what we want?".
I agree -- excellent questions.
I think I`d like to see an article with questions like this which lays out what answers to these questions are currently in place. There probably aren`t very many people who know the current answers. Such an article would also need to explain for each question why it is necessary to ask it now.
I'll think about it. :)
On 03/21/2014 10:13 AM, Matthew Miller wrote:
Sure it is. Again, hyperbole doesn't really help. In fact, in many situations, an MTA doesn't do you any good, as your network won't allow it to do anything useful remotely (true now at most big companies and on most big ISPs) and local delivery goes into the black hole of root's mailbox unless you configure it. And if you're going to configure one, installing one isn't a significant extra step. Plus, since there are multiple different MTAs, this is back to your choice of fvwm vs. gnome -- if we have a default you don't want, you're going have to remove it to put in your choice.
If you do need to use sendmail, and your ISP is blocking Port 25, it's not that hard to configure things to use a smarthost. As an example, I have my own (vanity) domain and use its mail servers, over Port 587. I also have sendmail configured to use that server and port, along with the appropriate username/password. If anybody out there needs to do the same thing, instructions are at http://www.zeff.us/SMTPAuth.txt
On Fri, Mar 21, 2014 at 10:27:02AM -0700, Joe Zeff wrote:
If you do need to use sendmail, and your ISP is blocking Port 25, it's not that hard to configure things to use a smarthost. As an example, I have my own (vanity) domain and use its mail servers, over Port 587. I also have sendmail configured to use that server and port, along with the appropriate username/password. If anybody out there needs to do the same thing, instructions are at http://www.zeff.us/SMTPAuth.txt
Absolutely. But since you need to configure it before it's useful, it's arguably actively harmful to have it running by default. That's all. No one is removing MTAs from the distro.
On 19 March 2014 15:53, Matthew Miller mattdm@fedoraproject.org wrote:
I promised a while ago that I would provide a text version of my talk at DevConf, for people who couldn't make it and because sitting through a video of me standing up there going on and on doesn't really make for good followup discussion.
Because it grew rather long, I think it works best as a web article, which you can find on Fedora Magazine at http://fedoramagazine.org/?p=1236.
Also because of the length, I'm posting it installments. This part is the background, next part is some of the things we are doing, followed by the panel discussion, followed by Q&A. Whew; that's a lot.
I can post the text here if people think it'd be really helpful to do so, (although I'm inclined to think that would be unwieldy), and in any case I will take questions, comments, complaints, in any media including replies here, on the article, on the social media, or at any bar or coffee shop within walking distance of Boston's MBTA.
Interesting... thanks for that.
Question. It seems implicit in the piece that Ubuntu has the eyeballs, as it were, even if it's not cool any more either.
Do you guys ever ask yourselves /why/ that is and if there's anything you could do to change it?
On 03/21/2014 10:30 AM, Matthew Miller wrote:
Absolutely. But since you need to configure it before it's useful, it's arguably actively harmful to have it running by default. That's all. No one is removing MTAs from the distro.
That depends. If your ISP isn't doing any port blocking, the default configuration is fine. The only way it can cause problems is if you need to configure it and haven't, and that's no worse than what happens if you need it and don't know it wasn't installed. What cases are you thinking of that would make it "actively harmful?"
On 03/21/2014 10:30 AM, Matthew Miller issued this missive:
On Fri, Mar 21, 2014 at 10:27:02AM -0700, Joe Zeff wrote:
If you do need to use sendmail, and your ISP is blocking Port 25, it's not that hard to configure things to use a smarthost. As an example, I have my own (vanity) domain and use its mail servers, over Port 587. I also have sendmail configured to use that server and port, along with the appropriate username/password. If anybody out there needs to do the same thing, instructions are at http://www.zeff.us/SMTPAuth.txt
Absolutely. But since you need to configure it before it's useful, it's arguably actively harmful to have it running by default. That's all. No one is removing MTAs from the distro.
I agree that an MTA needs to be installed unless you're installing a "server" (and Fedora really doesn't offer that model anymore).
For quite a while, Fedora had installed an MTA and started it. By default, it was configured to only listen to localhost on port 25 so it was more-or-less innocuous. The fact that F19 stopped installing an MTA by default caught a lot of people off guard.
Unlike others, however, I find the new system logging and analysis tools cumbersome and painful to use. Having a program send an email to me if it encounters issues is FAR superior to me having to plow through the logs to see if it ran correctly or not. ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks@alldigital.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - ----------------------------------------------------------------------
On Fri, Mar 21, 2014 at 05:50:48PM +0000, Liam Proven wrote:
Interesting... thanks for that. Question. It seems implicit in the piece that Ubuntu has the eyeballs, as it were, even if it's not cool any more either. Do you guys ever ask yourselves /why/ that is and if there's anything you could do to change it?
Sure. :)
And I think we know some of the answers, including taking advantage of the gap between RHEL's long enterprise life and Fedora's very short one, targeting polish on the desktop, and excellent community building and marketing. We can do some of those things, but not necessarily in the same way, and I don't think we'll be able to do anything meaningful by imitating what they did years ago -- that window is past and we need to go our own way.
On Fri, Mar 21, 2014 at 10:55:56AM -0700, Rick Stevens wrote:
Unlike others, however, I find the new system logging and analysis tools cumbersome and painful to use. Having a program send an email to me if it encounters issues is FAR superior to me having to plow through the logs to see if it ran correctly or not.
We definitely need better alerting and monitoring, mail-based or otherwise. Help wanted. :)
On 03/21/2014 10:50 AM, Liam Proven wrote:
Question. It seems implicit in the piece that Ubuntu has the eyeballs, as it were, even if it's not cool any more either.
Do you guys ever ask yourselves/why/ that is and if there's anything you could do to change it?
I'm guessing that you mean that Ubuntu is more popular, and has a much bigger installed user base. If so, it's probably because it's designed to be very user friendly, doesn't make a big deal about some of the software restrictions that Fedora cares about and is, as I like to say, designed for "Windows refugees." Fedora, OTOH, is a much more geeky distro designed as a test bed for new ideas, programs and technologies that's not for people who don't like to tinker with things or who aren't willing to accept that not everything in their distro is really ready for prime time.
Candidly, I'm expecting zorin (http://zorin-os.com/) to become a significant part of the Linux world once Microsoft finally drives a stake through XP's heart because it's a fork of Ubuntu designed to have a UI that looks as much like XP as possible, so that people can pretend they're still using Windows.
On Fri, 21 Mar 2014 13:13:28 -0400 Matthew Miller wrote:
Saying something "sucks" isn't very helpful. Not only is it needlessly negative, it is intangible. Name a real problem and we can talk about it.
Every single aspect of it is a real problem. In the dictionary they have a picture of it next to the word "disaster". The whole random wheel/spoke thing is an invitation to forget to do something important. And who the hell would think that "network" would be the spoke where you set the system name? Why is the screen full of blank space and cryptic meaningless icons when it could use some of that space for valuable hints in actual text you can read?
On the specific you do give, I'm pretty confident in saying that you're actually wrong.
And therein lies the problem. Everyone who worked on partitioning has been congratulating each other for so long you can't see any problems. The partitioning is actually absolutely impossible to use:
If there are multiple disks, it provides you only the size, model, and serial number to distinguish them, as though everyone has that memorized. Your only actually choice is to pick them all and hope you can see a clue later.
After picking which disks to partition, you are presented with a single choice: You can press the Done button even though you are not anywhere close to done, or you can say "Screw this" and install a different distro.
Should you work up the courage to press the Done button, you then reach the next layer of total obfuscation.
It names the partitions by what operating systems are installed on them. Who the hell thinks that way? And where do I look for partitions that don't have any operating system on them. Why not name them by the biggest video file that lives on the partition or some other random choice?
The adjectives "intuitive" or "useful" cannot be applied to any part of this interface. It is absolute and total junk.
Our only hope is for redhat to use this in the next RHEL and then we can probably get changes as wave after wave of enterprise users descend on redhat HQ with pitchforks and torches.
On Fri, Mar 21, 2014 at 11:04:24AM -0700, Joe Zeff wrote:
I'm guessing that you mean that Ubuntu is more popular, and has a much bigger installed user base. If so, it's probably because it's designed to be very user friendly, doesn't make a big deal about some of the software restrictions that Fedora cares about and is, as I like to say, designed for "Windows refugees." Fedora, OTOH, is a much more geeky distro designed as a test bed for new ideas, programs and technologies that's not for people who don't like to tinker with things or who aren't willing to accept that not everything in their distro is really ready for prime time.
+1 to this, too. I don't think we *want* to become the mainstream windows-replacement OS, because that's not where we want to be on the innovation curve.
On 21 March 2014 18:04, Joe Zeff joe@zeff.us wrote:
I'm guessing that you mean that Ubuntu is more popular, and has a much bigger installed user base.
Yes, that's right.
If so, it's probably because it's designed to be very user friendly,
True, I think it succeeds better there. I have been running Ubuntu since the first release, on a huge variety of hardware, including some non-x86 kit.
To be brutally honest, even with commercial live-deployment Unix experience going back to 1989, I have /never/ got any release of Fedora to successfully install on real hardware. I find the installer that bad, and it's been getting worse, not better.
doesn't make a big deal about some of the software restrictions that Fedora cares about
That's a fair point, but other distros cover that base too - Debian, primarily. I could be wrong but I don't see that as a perceived Fedora strength.
and is, as I like to say, designed for "Windows refugees."
It's designed to be easy, sure - that's the reason for the "Linux for Human Beings" strapline. It's not /remotely/ Windows-like, though, and never really particularly has been.
Fedora, OTOH, is a much more geeky distro
Hmmm. I don't see that either. Not saying you are wrong, just I have not got that message.
To me, the scale of hardcore geeky distros goes something like (less -> more):
Debian -> Slackware -> Arch -> Gentoo
designed as a test bed for new ideas, programs and technologies
That's the primary message, yes.
that's not for people who don't like to tinker with things or who aren't willing to accept that not everything in their distro is really ready for prime time.
[Nod] Yup.
Candidly, I'm expecting zorin (http://zorin-os.com/) to become a significant part of the Linux world once Microsoft finally drives a stake through XP's heart because it's a fork of Ubuntu designed to have a UI that looks as much like XP as possible, so that people can pretend they're still using Windows.
I have been looking at Zorin for a possible review recently. I wasn't that impressed. There are /loads/ of Windows-like Ubuntu remixes already - #1 is Mint, then there's Lubuntu; Xubuntu can be made into it trivially easily. Zorin is no more Windows-like than any of them.
Every single aspect of it is a real problem. In the dictionary they have a picture of it next to the word "disaster". The whole random wheel/spoke thing is an invitation to forget to do something important. And who the hell would think that "network" would be the spoke where you set the system name? Why is the screen full of blank space and cryptic meaningless icons when it could use some of that space for valuable hints in actual text you can read?
The system name or hostname is important to networking; so, I can see why it's under networking, but I believe your frustration is more related to the lack of guidance and quality than anything else.
The old installer did a much better job of 'guiding' the user through a set path of installation which achieved greater visibility and made it hard to skip or miss options. Of course, the negative side was the user could have an overwhelming feeling due to the tremendous amount of time it took to go through everything. The new installer sacrifices visibility and guidance for a more free process that sometimes has a very cookie-cutter or blank feeling to every screen, but it drastically reduces the time spent in installation since the user can bounce around wherever they want.
Matthew Miller wrote:
On the specific you do give, I'm pretty confident in saying that you're actually wrong.
And therein lies the problem. Everyone who worked on partitioning has been congratulating each other for so long you can't see any problems. The partitioning is actually absolutely impossible to use:
While I agree that Matthew's response was naive, you didn't offer any initial feedback that supported your feelings.
If there are multiple disks, it provides you only the size, model, and serial number to distinguish them, as though everyone has that memorized. Your only actually choice is to pick them all and hope you can see a clue later.
I believe there is a link towards the bottom left that will show more details, but it's easy to miss.
After picking which disks to partition, you are presented with a single choice: You can press the Done button even though you are not anywhere close to done, or you can say "Screw this" and install a different distro.
Should you work up the courage to press the Done button, you then reach the next layer of total obfuscation.
I agree that there is a usability issue. The installer almost has the feeling that it was designed to be the equivalent of a senseless click fest like installing adware bundled with software. I guess that is sort of what they wanted... quicker experience, but I don't think they've gotten the right mixture.
The installation team needs to work on bridging the gap a little... and I think they will.
The adjectives "intuitive" or "useful" cannot be applied to any part of this interface. It is absolute and total junk.
Our only hope is for redhat to use this in the next RHEL and then we can probably get changes as wave after wave of enterprise users descend on redhat HQ with pitchforks and torches.
I know you're frustrated, but this is what Fedora was designed for: a more up-to-date distro that can iterate on new features to get them right before adding it to RHEL. Just keep your fingers crossed that they fix a few issues before Fedora 21 now that they're aware of your complaints :-)
On Fri, Mar 21, 2014 at 07:14:57PM +0000, Powell, Michael wrote:
The old installer did a much better job of 'guiding' the user through a set path of installation which achieved greater visibility and made it hard to skip or miss options. Of course, the negative side was the user could have an overwhelming feeling due to the tremendous amount of time it took to go through everything. The new installer sacrifices visibility and guidance for a more free process that sometimes has a very cookie-cutter or blank feeling to every screen, but it drastically reduces the time spent in installation since the user can bounce around wherever they want.
I don't get it, how is "bouncing around" faster? I bounced around for an eternity when I installed F20 on my new laptop. Earlier it used to take me something of the order of an hour to get everything done, this time around it was so frustrating I gave up after a couple of hours more than once, finally one weekend I decided I get this done or I do not sleep.
More information is not a problem, if it is something that is non-essential, an "optional" label next to it should be enough. This time around my main problem was anaconda just would not let me choose the partition sizes I wanted. In the end I went with putting in a skeleton scheme that would let me install Fedora, and alter the partitioning as I wanted post-install. This is by far the worst disk partitioning interface in a Fedora installer I have used since F10 (I started using Fedora regularly then).
I don't get it, how is "bouncing around" faster? I bounced around for an eternity when I installed F20 on my new laptop. Earlier it used to take me something of the order of an hour to get everything done, this time around it was so frustrating I gave up after a couple of hours more than once, finally one weekend I decided I get this done or I do not sleep.
Absolute ease of use is what they're going for and the easiest way to achieve that is to reduce the amount of information a user has to process.
Bouncing around or free-form favors hiding details and relying on defaults; it's not ideal for every user, but it does lessen the information to process and thereby the time. By comparison, the old installer forced users that relied on defaults to process almost as many screens as someone that wanted to customize everything.
I don't think they have the right balance, but I think they can fix it...
More information is not a problem, if it is something that is non-essential, an "optional" label next to it should be enough. This time around my main problem was anaconda just would not let me choose the partition sizes I wanted. In the end I went with putting in a skeleton scheme that would let me install Fedora, and alter the partitioning as I wanted post-install. This is by far the worst disk partitioning interface in a Fedora installer I have used since F10 (I started using Fedora regularly then).
I completely agree that the partitioning UI needs work...
On Mar 21, 2014, at 12:10 PM, Tom Horsley horsley1953@gmail.com wrote:
On Fri, 21 Mar 2014 13:13:28 -0400 Matthew Miller wrote:
Saying something "sucks" isn't very helpful. Not only is it needlessly negative, it is intangible. Name a real problem and we can talk about it.
Every single aspect of it is a real problem. In the dictionary they have a picture of it next to the word "disaster". The whole random wheel/spoke thing is an invitation to forget to do something important. And who the hell would think that "network" would be the spoke where you set the system name?
Where else would you put it? It has nothing to do with date/time or destination or source. I associate it with network because hostname->Avahi-->network.
Why is the screen full of blank space and cryptic meaningless icons when it could use some of that space for valuable hints in actual text you can read?
Right the problem is that it's not wordy enough (lovely to see much wordiness going away in Rawhide at this point). More tool tips might be nice.
On the specific you do give, I'm pretty confident in saying that you're actually wrong.
And therein lies the problem. Everyone who worked on partitioning has been congratulating each other for so long you can't see any problems.
? It's only been ~15 months since it was publicly released, that's long?
The partitioning is actually absolutely impossible to use:
Ahh, so when hyperbole simply isn't going far enough we actually have to descend into the obviously ridiculous, as in, worthy of ridicule. My house plant can do a Fedora install with this installer. That you keep failing to get any kind of successful installation is a bit amusing. Maybe you need more water?
If there are multiple disks, it provides you only the size, model, and serial number to distinguish them, as though everyone has that memorized. Your only actually choice is to pick them all and hope you can see a clue later.
Yeah this is a weak area, but I'm not sure if there are bugs filed so feel free to check and file one and report back.
The UI probably ought to show volume names first, partition names second. I think it's a separate UI floating panel to show a subset of lsblk + blkid. It's fairly sloppy to not use volume or partition labels these days though, so you pretty much get what you're asking for if you're not using them already. Fortunately, anaconda does apply a volume label during install these days. Other distros don't leverage it.
Although not user friendly for a GUI, you could switch to a shell and do lsblk and blkid and figure things out without having to choose all disks.
After picking which disks to partition, you are presented with a single choice: You can press the Done button even though you are not anywhere close to done, or you can say "Screw this" and install a different distro.
Haha. That's like going to the grocery store, getting 1/2 the stuff you need, finally becoming exasperated at that point, saying "screw this" and going fishing instead. That's a hilarious reaction.
Don't even bother with a "hey I'm confused" or a "hey wtfsauce" on any user list or forum. Just go straight to abandoning the path. "I felt like a nice cup of Earl Grey and scones today, but I have to wait in line, so SCREW THIS, I'm going to move out of my apartment instead."
Wow ok…
Hey, your right eye is REALLY twitching a lot, you might need to lay off the coffee after, say, 8am.
Should you work up the courage to press the Done button, you then reach the next layer of total obfuscation.
Right. The hub! It's a completely different layer you've never seen before, if you have amnesia. Oh let's just go ahead and add that to the list of anaconda gripes….
c a u s e s a m n e s i a
It names the partitions by what operating systems are installed on them.
You are confused. First, MBR partition scheme doesn't support partition names. Second, GPT does, but the only partition I've seen named in a Fedora install is "EFI System Partition", the others are unnamed. I'm not sure what utility actually names it.
What you are thinking of are the volume labels, and you can change them in the UI, it's just a default.
Who the hell thinks that way? And where do I look for partitions that don't have any operating system on them.
Right who the hell thinks about a default volume label after the name/version of the distribution? That's just crazy! I don't know, let's try one even better like name them the biggest video file on the system… that's not going to cause ANY trouble at all. *wink* *wink*
AlcoholCommercial
Hahaha, that's funny at least. Not helpful in ID'ing the drive, at all. But funny!
Why not name them by the biggest video file that lives on the partition or some other random choice?
Hmm that's quite an obscure game. Let's try that with cities based on building sizes:
Republic Plaza Columbia Center Toranomon Hills Company Business Towers Torre A Post Tower
I almost feel like continuing. I'm learning things I didn't know, things I'll never remember. I will bet you can't even name the continents those are on, let alone the cities.
I've heard a lot of bad ideas today, but this is definitely, DEFINITELY #2. Just kidding, it's #1.
The adjectives "intuitive" or "useful" cannot be applied to any part of this interface. It is absolute and total junk.
I know exactly what you mean by this, I had one hell of a time figuring out what the ballz the Time & Date spoke was all about. I nearly gave myself an aneurism trying to figure it out.
Our only hope is for redhat to use this in the next RHEL and then we can probably get changes as wave after wave of enterprise users descend on redhat HQ with pitchforks and torches.
Right because most RHEL installs aren't kickstart installs, they're GUI installs. And enterprise customers have so much spare time on their hands after paying for RHEL service/support that the thing they use the least is the casus belli to go out to the woodshed for pitchforks and torches. Riiiight.
Right now, I'm looking at a bag full of nuts. (I'm not kidding it's a LITERAL bag full of nuts.) Where is poma? I'm certain he'll have a better response than this. Oh wait he's already done it!! BRILLIANT! He didn't respond! Pure genius!
Chris Murphy
On 21 March 2014 23:25, Chris Murphy lists@colorremedies.com wrote:
Ahh, so when hyperbole simply isn't going far enough we actually have to descend into the obviously ridiculous, as in, worthy of ridicule. My house plant can do a Fedora install with this installer. That you keep failing to get any kind of successful installation is a bit amusing. Maybe you need more water?
Firstly, your mocking hectoring tone is very unhelpful, annoying and is not a productive way to engage.
Secondly, I can confirm this finding. I was completely unable to install F20 using the current installer program. My system has 2 drives - a 1TB HD and a 120GB SSD. The SSD holds Ubuntu and Win7; the 1TB drive holds /home, swap, a dedicated Windows swap partition, a Windows data drive, and 2 spare unused root partitions for test distros.
This is /not/ a very complex layout - there is no RAID, no LVM, no GPT, nothing hairy or difficult.
The F20 installer was completely unable to understand it and allow me to install a complete system. Assigned some 250GiB of space, it said that it needed 6.5GB and there wasn't enough room.
In trying to install, it erased one of the spare-root partitions and was unable to recreate it in the available empty space.
It *is* broken and it *is* unusable. "Well it works for me" is *not* an adequate reply.
On Mar 21, 2014, at 6:18 PM, Liam Proven lproven@gmail.com wrote:
On 21 March 2014 23:25, Chris Murphy lists@colorremedies.com wrote:
Ahh, so when hyperbole simply isn't going far enough we actually have to descend into the obviously ridiculous, as in, worthy of ridicule. My house plant can do a Fedora install with this installer. That you keep failing to get any kind of successful installation is a bit amusing. Maybe you need more water?
Firstly, your mocking hectoring tone is very unhelpful, annoying and is not a productive way to engage.
Are you asserting that people who bitch and whine about the installer are entitled to a monopoly on mocking hectoring tone; and unhelpful, annoying, unproductive engagement?
Secondly, I can confirm this finding. I was completely unable to install F20 using the current installer program. My system has 2 drives - a 1TB HD and a 120GB SSD. The SSD holds Ubuntu and Win7; the 1TB drive holds /home, swap, a dedicated Windows swap partition, a Windows data drive, and 2 spare unused root partitions for test distros.
This is /not/ a very complex layout - there is no RAID, no LVM, no GPT, nothing hairy or difficult.
It is actually a complex layout. Most of the world's installers can't deal with what you just described. The #1 OS install today is software restore: point to a physical device (assuming it even supports more than one which many don't), and it obliterates all data on the drive, partitions it in a predefined way, installs predefined software. You have no control other than the binary condition: do or do not.
The #2 OS install is the Windows retail/update installer. It mainly expects to install to a blank drive, but has an advanced mode that permits rudimentary partitioning, including partition deletion and install to free space. But no resizing. #3 OS install is OS X, it only permits the selection of one pre-formatted volume, you have to use a separate utility if you want to do any partitioning.
And yes, it's fair to bring up what money bags companies with more money and resources than god. Because this is an area where they've all considered what you're describing, is an edge case. Not common at all.
The Fedora 20 installer's default/easy/guided/auto path installs to free space. Yet it has more options and outcomes than the total number of all possible options in both the Windows and OS X installers combined.
Hmm. Now I believe you were just about to cite a bugzilla ID describing the above behavior?
The F20 installer was completely unable to understand it and allow me to install a complete system. Assigned some 250GiB of space, it said that it needed 6.5GB and there wasn't enough room.
I've done hundreds of hours of installer testing over the last year. It has been really frustrating. This is the most complicated/capable installer I've ever worked with other than maybe the OpenSUSE installer. Out of the gate it offerred too much compared to the time/resources allotted for QA, debugging, and code changes needed.
The reality is, you get either stability or you get features. You don't get both. The mantra for the new installer was about getting as many of old installer's features into the new one as soon as possible, and stability was simply expected to have to take a hit in order to do that. And that's exactly what happened.
Let's pretend the installer could only do 20% of what the old installer did, yet it was almost bullet proof - never crashed, didn't have any of the logic problems you're talking about, and so on. Would Fedora users have understood that trade off? Maybe a lot of them would have. But then we'd have a lot of others pining for a right to a GUI that lets them create some of the most esoteric storage layouts of all time.
And guess what? That has to be coded, and ostensibly should be tested. And quite frankly the QA resources are really limited. Not every possible combination permitted in Manual Partitioning is tested at all. That's how much it can do. It's nearly unlimited possibilities because, guess another thing, I've never once seen it disqualify a drive layout from the start. I've never seen it look at a crazy layout and go "umm yeah, no please use gparted and obliterate this drive first." But I've seen that many times with the OS X installer: flat out refusal, "go format the drive in Disk Utility." Quite a few times when trying to prepare a drive for dual boot on OS X I've seen the error message that the disk can't be partitioned, and that I had to obliterate the whole drive and reinstall OS X from scratch in order to install Windows side by side. So really, anaconda is extremely tolerant and I think that's something of a problem too. It probably should be disqualifying a lot of nut case layouts, and just saying no.
In trying to install, it erased one of the spare-root partitions and was unable to recreate it in the available empty space.
And you have a bug for this? It's *really* difficult to get the installer to inadvertently delete partitions. It requires two clicks: selection, then deletion. For guided partitioning, the button is labeled "delete" whereas the button in manual partitioning is labeled as a minus symbol.
One thing that some people don't easily grok is that Manual Partitioning isn't partitioning oriented. It's mount point oriented. And that's because mount points can be partitions, subvolumes, logical volumes, or md block devices. It's not correct to call all of those things partitions. But all of them ultimately are assigned mount points. This is a top down view, rather than the typical bottom up view where you always have partitions, and then maybe you have raid devices or LVs or subvolumes. The idea is to think less about the details of the layout and more about the outcome you want.
This is understandably confusing if you're really familiar with storage stack creation. But most people aren't. Nevertheless, one of the first steps is drive selection, which is about the most bottom layer there is. And then the very next step is the top most layer, which are the mount points. So it's an unexpected context shift from bottom to top, seemingly without any conversation about what's happening in the middle. It's a different approach.
If you remain attached that what you're doing in Manual Partitioning is in fact partitioning, you'll continue to be frustrated.
It *is* broken and it *is* unusable. "Well it works for me" is *not* an adequate reply.
I didn't say "well it works for me" I said it's ridiculous to say it's impossible to use. And I'll partly walk that back because if you're going to be entitled to stubborness, and unwilling to adapt to the layouts you can have rather than the layout you want, then yes you might be hitting a brick wall. But I assure you that the vast majority of the world's installers would have poo poo'd you much sooner.
Now, if you didn't file a bug about your anecdote, I want you to imagine me staring at you with a look of "really?" Because this much effort complaining yet no bug report? How exactly do you expect it to get better?
Chris Murphy
On 03/22/14 08:18, Liam Proven wrote:
Secondly, I can confirm this finding. I was completely unable to install F20 using the current installer program. My system has 2 drives - a 1TB HD and a 120GB SSD. The SSD holds Ubuntu and Win7; the 1TB drive holds /home, swap, a dedicated Windows swap partition, a Windows data drive, and 2 spare unused root partitions for test distros.
I haven't had the need for multi-boot systems since I use VM's. But your post got me curious.
I had a Win7 VM with a single disk fully used with NTFS file systems, I'm not using a dedicated Windows Swap, and I did the following.......
1. Cloned the Win7 system and added a second disk. 2. Installed Ubuntu on the second drive using a portion of the space and standard partitioning. 3. Rebooted and tested that I could access Win7 and Ubuntu systems. 4. Booted into Ubuntu and used parted to create dummy unused partitions. 5. Then did the F20 installation and selected dummy partitions for reclamation and the installation completed.
My only "complaint" is that when I rebooted there was no menu entry for Fedora. Apparently the install process wrote the boot-loader to the second disk as that was the only one I had "selected". I missed the link appearing in the lower left about the boot-loader otherwise I would have caught it. I wish that link/option would have been more prominent/conspicuous.
So, I re-did step #5 this time selecting both disks and again reclaiming the space that F20 was now taking up.
On reboot I had options to boot F20, F20 rescue, Win7, Ubuntu, and Ubuntu with options.
And all is well.....
So, I don't know what may be different in my configuration (maybe the windows swap?) that allowed me to compete the exercise. But surely, it isn't totally broken for everyone.
So, I don't know what may be different in my configuration (maybe the windows swap?) that allowed me to compete the exercise. But surely, it isn't totally broken for everyone.
Perhaps not. And in fact, I have sort of figured out what to do to get exactly the old anaconda way of doing things for me. (I say sort of.) The trouble is that lots and lots of stuff is hidden in nondescript and (to many, a counterintuitive) way of doing things.
The problem is that we don't reinstall (or install F) every few weeks or months so it is very easy to forget what/how this was done every six months. Yes, we can take notes, but recall that this is a time when making notes on the machine itself is not that easy because we are at the installation phase. Also, many things have to be discovered and in particular getting to the manual paritioning setup to me was completely hard to follow: you hit Done or something like that and then you get to the installer.
Months later, I have figured out that there are buttons to set the time according to a desired network provided. It is all hidden, and I accidentally found it by looking and wondering why this option should not have been included in the install options.
I guess some of us are trying to say that the installer has become quite opaque. Some of us can scrape away the opacity easily sometimes, some can do it rarely, while some can do it all the time. The question that I wonder about is: was this at all necessary.
Personally, I like letting stable functional things stay, and think that it would have been better to rebuild on the old installer. But then I am not at all an expert in any of this, and maybe this was not possible. However, the new installer is far too removed in process from the old one, and the least that could be done could be to include a manual and an FAQ (question bank) with the installer. This could be searchable (and included with the installer, because, again note that the machines functionality is not that high at the time of the install). Of course, this takes effort and time which volunteer developers may not have or wish to put into.
Btw, the new installer is uglier in terms of looks than the old, too, but that is a matter of personal like!
My rambling is all to say that while I eventually got my stuff the way I wanted, I am not sure if it should have taken all that while. I had installs happening in 3-4 minutes with the old installer (now the time taken is exponentially higher). So, while I appreciate all the hard work done by the developers, I can not but wonder if their hard work has been in vain (to an extent) in the sense that many of us old and loyal F users are unable to fully appreciate all their contributions in this regard.
Best wishes, Ranjan
Allegedly, on or about 21 March 2014, Joe Zeff sent:
I'm guessing that you mean that Ubuntu is more popular, and has a much bigger installed user base. If so, it's probably because it's designed to be very user friendly, doesn't make a big deal about some of the software restrictions that Fedora cares about and is, as I like to say, designed for "Windows refugees."
I see no point in all Linux OSs trying to be /that/, let Ubuntu fill that role, and Fedora fill its own. Nor is their any point in all Linuxes being the same as each other. I use Linux, and Fedora Linux, because I don't want to use Windows. I don't want anything that's like it, in behaviour, nor looks.
People seem to harp on about success by being the leader of the bunch, and only that measure. Whereas being successful is really being able to carry on doing what you want to. If Fedora can be Fedora, and people want to use it, then it's succeeded. It's not a commercial product, don't judge it by commercial measures.
Candidly, I'm expecting zorin (http://zorin-os.com/) to become a significant part of the Linux world once Microsoft finally drives a stake through XP's heart because it's a fork of Ubuntu designed to have a UI that looks as much like XP as possible, so that people can pretend they're still using Windows.
With a name like that, I half expect it to be foisted upon as by Christopher Walken (shudder)... Google "Max Zorin," if you don't know your super evil James Bond villians.
Allegedly, on or about 22 March 2014, Ranjan Maitra sent:
Months later, I have figured out that there are buttons to set the time according to a desired network provided. It is all hidden, and I accidentally found it by looking and wondering why this option should not have been included in the install options.
It used to be.
Allegedly, on or about 21 March 2014, Powell, Michael sent:
The system name or hostname is important to networking; so, I can see why it's under networking, but I believe your frustration is more related to the lack of guidance and quality than anything else.
Well, actually, for a lot of people, the system name is simply what they want to call the computer. The computer may not even be on a network, at all. There's certainly cause for having a process of naming the computer.
For LANs, the hostname may not actually be set on the computer. A DHCP server may dole out IPs and hostnames, or simply dole out hostnames and a DNS lookup discovers what name is given to that IP. In that scenario there's some logic to using the network configuration to fill in the name. Though there's still the reverse case, where I want to name a computer, and let the rest of the network find out what my name is, from me (well, /this/ computer, not actually me, personally).
The old installer did a much better job of 'guiding' the user through a set path of installation which achieved greater visibility and made it hard to skip or miss options. Of course, the negative side was the user could have an overwhelming feeling due to the tremendous amount of time it took to go through everything. The new installer sacrifices visibility and guidance for a more free process that sometimes has a very cookie-cutter or blank feeling to every screen, but it drastically reduces the time spent in installation since the user can bounce around wherever they want.
I found the old installer (e.g Fedora 9 vintage) to be quite basic in what you had to play with. Sure there were options you could fiddle with. But you could simply opt for one of about four installation types (general desktop, minimal install, etc.). You could opt for fiddling with partitions, or let it erase and install. I didn't find it hard to manually set up partitions. The set and forget approach worked fairly well to get the install run. And the post-install routine was fairly simple, too (basic networking, set the clock, etc.).
On 22 March 2014 12:04, Tim ignored_mailbox@yahoo.com.au wrote:
I see no point in all Linux OSs trying to be /that/, let Ubuntu fill that role, and Fedora fill its own. Nor is their any point in all Linuxes being the same as each other. I use Linux, and Fedora Linux, because I don't want to use Windows. I don't want anything that's like it, in behaviour, nor looks.
Agreed. /But/ it's damned handy if something works on all hardware that Windows supports, understands Windows filesystems and network protocols and file formats and so on. Be careful not to throw out the baby with the bathwater.
People seem to harp on about success by being the leader of the bunch, and only that measure. Whereas being successful is really being able to carry on doing what you want to. If Fedora can be Fedora, and people want to use it, then it's succeeded. It's not a commercial product, don't judge it by commercial measures.
Agreed again - /but/ I think that there are questions which could be profitably asked.
E.g. some third-party .deb packages can be installed on various versions of Ubuntu, all the Ubuntu remixes official and unofficial, and even on Debian itself. This is not universally the case, but .deb always means apt or something on top of apt, such as aptitude, and .deb packages are very portable across the Debian universe.
The same is /not/ true of RPM. RHEL and Fedora use one meta-package-manager, SUSE uses a different one, Mandriva and Mageia use a different one again. You can't install SUSE packages on non-SUSE distros that use RPM, and you can't install RH RPMs on non-RH distros.
There is a lot of infighting among RPM distros - it probably is not perceived as such outside the RH world, but that is how it looks from elsewhere. There are even forks of RPM itself - e.g. http://rpm5.org/
I don't know but I bet that online repository formats are incompatible, too.
This kind of thing is an issue, and unless or until it's fixed, the Debian family will hold an upper hand there.
-- Liam Proven * Profile: http://lproven.livejournal.com/profile Email: lproven@cix.co.uk * GMail/G+/Twitter/Flickr/Facebook: lproven MSN: lproven@hotmail.com * Skype/AIM/Yahoo/LinkedIn: liamproven Tel: +44 20-8685-0498 * Cell: +44 7939-087884
Joe Zeff wrote:
Fedora, OTOH, is a much more geeky distro designed as a test bed for new ideas, programs and technologies that's not for people who don't like to tinker with things or who aren't willing to accept that not everything in their distro is really ready for prime time.
I'm afraid this "bleeding-edge" metaphor is often used as an excuse for missing or poor documentation. Official Fedora documentation is excessively verbose, and not good at answering - or even considering - issues and problems that users are likely to face. There is - or should be - a big difference between a Reference Manual and a User Manual.
As an example (at random) I'd take using a USB stick for installation. Does anybody nowadays actually burn CDs or DVDs? Yet finding instructions on the use of USB sticks (and external USB drives) in the official installation manual involves jumping around from chapter to chapter.
On 03/22/14 21:28, Timothy Murphy wrote:
Does anybody nowadays actually burn CDs or DVDs?
Yes..... The BIOS on many of the machines I encounter don't support booting from USB.
Firstly, your mocking hectoring tone is very unhelpful, annoying and is not a productive way to engage.
Are you asserting that people who bitch and whine about the installer are entitled to a monopoly on mocking hectoring tone; and unhelpful, annoying, unproductive engagement?
Hey all. I know it's fun to engage in hyperbolic mudslinging on a mailing list, but rather than have it escalate, let's please not do it on either side of the disagreement. It starts out in humor but gets out of control quickly and doesn't add to the discussion -- it certainly doesn't help convince anyone who disagrees, and it makes the whole conversation less useful overall. Thanks.
On 22 March 2014 03:54, Chris Murphy lists@colorremedies.com wrote:
Are you asserting that people who bitch and whine about the installer are entitled to a monopoly on mocking hectoring tone; and unhelpful, annoying, unproductive engagement?
I don't think any of us should do it, ideally. But perhaps in gentle chiding of someone who could have answered their question with Google in 10sec, then OK.
When someone is unable to even install the whole OS, no. Inappropriate.
It is actually a complex layout.
I beg to differ. It is not /trivial/ but it is not complete.
Most of the world's installers can't deal with what you just described.
Factually incorrect. Windows 7, Ubuntu 13.10, Debian 7, Crunchbang and Elementary OS all had no problems.
On my desktop PC, I have a similar layout with Windows 8, Mac OS X 10.6 and Ubuntu 13.10. Again, no problems at all.
The #1 OS install today is software restore [...]
In my extensive experience of OSes going back to when I entered the business in 1988, following about 6-7y as a hobbyist, this is incorrect. You're describing one OS, principally - modern Windows.
And yes, it's fair to bring up what money bags companies with more money and resources than god. Because this is an area where they've all considered what you're describing, is an edge case. Not common at all.
Let's put it this way. I review and evaluate software for a living and have done for about 18y now. I have been using Linux since 1996 and have reviewed something like 20 or 30 distributions in that time.
Fedora 20's is *the least flexible* and least-capable installer I have ever seen on any distro of any kind, including x86, 68k, PowerPC or SPARC hardware since 1996.
The Fedora 20 installer's default/easy/guided/auto path installs to free space. Yet it has more options and outcomes than the total number of all possible options in both the Windows and OS X installers combined.
That is not my direct personal experience. I can demonstrate what I mean with screenshots and comparative step-by-step walkthroughs.
Hmm. Now I believe you were just about to cite a bugzilla ID describing the above behavior?
No. Why should I? Total failure to install the OS leaves me unable to use its bug-tracking tools, if any.
FWIW, I also tried installing on a completely empty standalone 250GB USB hard disk. It failed on that, too - it hung after the process began and never recovered. After about 6 hours, I power-cycled the machine.
As I have said previously, I have /never/ successfully installed Fedora on actual hardware since v1.0 shipped in, what was it, 2003? I have installed Haiku, Aros, FreeBSD, PC BSD, dozens of Linux distros, Windows 2 through 8, SCO Xenix, SCO Unix, OpenSolaris, OpenVMS, FreeDOS, DR-DOS, MS-DOS, PC-DOS, OS/2 1 through eComStation 2, MacOS 6 through OS X 10.9. I am *not* a newbie and I am *not* an inexperienced inexpert fumbler.
I've only got Fedora running in VMs. Even Slackware is easier.
I've done hundreds of hours of installer testing over the last year. It has been really frustrating. This is the most complicated/capable installer I've ever worked with other than maybe the OpenSUSE installer. Out of the gate it offerred too much compared to the time/resources allotted for QA, debugging, and code changes needed.
I have to tell you that in my experience of approaching a hundred installers, it is about the least capable of any C21 OS I have ever seen. I think it might beat eComStation but nothing else.
The reality is, you get either stability or you get features. You don't get both.
Everyone else manages.
But I've seen that many times with the OS X installer: flat out refusal, "go format the drive in Disk Utility."
I'm happy to believe you. OS X is fussier. But I'll tell you what, in this room I have PowerMacs booting MacOS 9, 10.4, 10.5 and 10.4 Server and MorphOS. MorphOS was a complete pig - I had to use an Ubuntu boot CD and Drive Genius to get a disk config it could install to, but I managed it. Sharing with both 10.4 and 10.5 on the same drive.
Fedora - nope. Not a single one of my test machines.
CentOS, yes, although I've seen it fail too. Fedora, nope.
So really, anaconda is extremely tolerant
Not in my direct personal experience, no, it is not.
and I think that's something of a problem too. It probably should be disqualifying a lot of nut case layouts, and just saying no.
Personally, not on anyone's behalf - that is insane. If the config is legal, it should be able to handle anything it's pointed at where there is at least one partition with enough room, or enough empty space to create such a partition. No legal layout should be refused.
And you have a bug for this?
No, what I have is material for a really bad product review.
It's *really* difficult to get the installer to inadvertently delete partitions. It requires two clicks: selection, then deletion. For guided partitioning, the button is labeled "delete" whereas the button in manual partitioning is labeled as a minus symbol.
It didn't just randomly delete it. It formatted it, then refused to install on it; 16GB was apparently not enough. Then I could not re-select that partition. So I deleted it, but the installer wouldn't create a new one.
The idea is to think less about the details of the layout and more about the outcome you want.
TBH I don't care what the idea is or was if the installer can't cope with a PC happily running 4 OSes already.
This is understandably confusing if you're really familiar with storage stack creation. But most people aren't.
I hope that I have demonstrated that I am not unfamiliar.
If you remain attached that what you're doing in Manual Partitioning is in fact partitioning, you'll continue to be frustrated.
IOW if I want to create the partitions and then put Fedora in what I've pre-created, it won't do it?
Then it is broken.
I didn't say "well it works for me" I said it's ridiculous to say it's impossible to use.
Then I will say it again to try to make my point.
I found it impossible to install the OS on my machine. The installer could not understand my disk layout. Every other distro I've tried can, therefore I feel safe in saying it's F20's problem.
And I'll partly walk that back because if you're going to be entitled to stubborness, and unwilling to adapt to the layouts you can have rather than the layout you want, then yes you might be hitting a brick wall.
Yup. A brick wall of not running an OS that can't handle my multiboot scenario successfully shared by 3 other working OSes.
But I assure you that the vast majority of the world's installers would have poo poo'd you much sooner.
And again, no, you're wrong. Sorry, but you are and I can, if you wish, show you.
Now, if you didn't file a bug about your anecdote, I want you to imagine me staring at you with a look of "really?" Because this much effort complaining yet no bug report? How exactly do you expect it to get better?
I have the info I wanted for a comparative OS article, like this one from a year ago:
http://www.theregister.co.uk/Print/2013/04/26/xbuntu_round_up/
My job there was done.
Meantime, for further Fedora eval, it's going in a VirtualBox. Sad, but that's all it seems able to handle.
I'll try CentOS on the bare metal and see if that copes.
On 22 March 2014 05:00, Ed Greshko ed.greshko@greshko.com wrote:
So, I don't know what may be different in my configuration (maybe the windows swap?) that allowed me to compete the exercise. But surely, it isn't totally broken for everyone.
I am not for a moment saying that it /is/ broken for everyone. Just for me, in a moderately sophisticated but not massively difficult multiboot scenario with 3 other distros and Windows.
On Sat, 22 Mar 2014 16:40:36 +0000 Liam Proven wrote:
Meantime, for further Fedora eval, it's going in a VirtualBox. Sad, but that's all it seems able to handle.
Actually, that's the key to installing on disk in a sensibly created partition layout: Install in a virtual machine 1st, then copy the VM disk image to a real disk and fiddle with the grub config and fstab. I install that way all the time now since the new installer appeared, and find it actually make things lots more convenient since I don't have to have the machine down while I'm doing the install.
On Mar 22, 2014, at 12:10 AM, Ranjan Maitra maitra.mbox.ignored@inbox.com wrote:
The problem is that we don't reinstall (or install F) every few weeks or months so it is very easy to forget what/how this was done every six months.
I think it'd be really helpful for users who care about the install experience, but don't use the installer often, contribute their experience. These users definitely have an insight that users more familiar with the installer don't have.
And I also think any application that irritates/frustrates is a bad UX and a legitimate gripe. So just because ultimately you can contort yourself to get the installer to do what you want, if that contortion annoyed you, there's a legitimate bug (or flawed design) somewhere. Good software shouldn't actively piss users off.
But a useful bug report still articulates why annoyance happened: usually a betrayal happened. User expects X, but software produced Y. So I'd write the bug in those terms. It's a UX bug.
A new behavior in Rawhide that's taken a lot of work is much better consistency in handling units. However as a side effect of using MiB and GiB, if you specify mb you get a conversion. If I create /boot with size 500 or 500m, it's created as 500 MiB. However if I specify size 500mb, it's created as size 476 MiB. Oh dear. That's technical correct, it's exactly what I asked for. But I think in storage the distinction of 2% between M and Mi, G and Gi is not helpful. The ship sailed a long time ago that base 1000 is used for storage, and base 1024 for memory. And then the next mind bender is when I do an autolayout and swap is 819.19 MiB on the left UI, but if I click that mount point, the detail right side UI Desired Capacity reports 819.1999998092651367/ and is then cut off, but if I scroll within that field to the right the full value is 819.19999980926513671875 MiB. In other words, it is actually possible to specify fractions of bytes in the storage UI. *sigh*
I'd like to see MB rounded to integers and GB rounded to one decimal place except in the lower left Available Space and Total Space fields, which should also be integers.
Yes, we can take notes, but recall that this is a time when making notes on the machine itself is not that easy because we are at the installation phase. Also, many things have to be discovered and in particular getting to the manual paritioning setup to me was completely hard to follow: you hit Done or something like that and then you get to the installer.
Oh right. The first Done button is what Liam was probably referring to, not the second Done button. Two Done buttons in the same spoke = UI failure. Ostensibly the UI's this is modeled after never ever do this. When their "upper left master button" element, the user is always always always returned to the hub. Here, Anaconda turns into a Which Way Book ((a.k.a. choose your own adventure book).
First, the spoke is titled Installation Destination. The first window upon entering that spoke is titled Installation Destination. The Done button very clearly indicates clicking it will return the user to the hub. But it never does. Huge betrayal of the UI. Instead you get a floating dialog, from which you choose from myriad options and either get another floating dialog or a whole separate page, wizard style, which also contains a Done button. Where the F am I? OK am I really Done now this time? Or are you just going to tease me again? OH I see this time we really are in fact Done.
(Just like the betrayals in Which Way Books. Except worse because there's no back button, like sticking your finger on the most recent choice page so that if you die you can go back and make a new choice, YES CHEATING. If I wish for a cheat finger in a UI, it has failed.)
In Rawhide this is a whole lot better in that you clearly, up front, choose two paths: automatically vs i will configure (which still confusing ends up being called Manual Partitioning it's technically no longer customize!). You still have the Done button that doesn't take you to the hub, therefore by definition it's actually a Next button wrongly named. (No, there's no pass for suggesting this Done button means "done with this page" because that's not how this type of UI is ever used anywhere else, even in Anaconda. It's supposed to be done with this spoke return to hub, but see I'm in a wizard UI now and wasn't properly informed.)
Anyway, it's better in Rawhide. It still has problems. There isn't a great way out of this problem. OK I've thought about it enough I filed a bug.
https://bugzilla.redhat.com/show_bug.cgi?id=1079655
Months later, I have figured out that there are buttons to set the time according to a desired network provided. It is all hidden, and I accidentally found it by looking and wondering why this option should not have been included in the install options.
I don't know what this refers to. Buttons to set time per desired network?
I guess some of us are trying to say that the installer has become quite opaque. Some of us can scrape away the opacity easily sometimes, some can do it rarely, while some can do it all the time. The question that I wonder about is: was this at all necessary.
I don't know what you mean by this. The installer rewrite? The installer UI big picture design? Specific UI elements? Obviously the installer team felt like a rewrite was needed to be more flexible in handling upcoming storage technologies like bcache, btrfs, and LVM thinp. As it turns out it's also more flexible being customized for Fedora products too, which wouldn't have been easy or maybe impossible with the previous installer.
Personally, I like letting stable functional things stay, and think that it would have been better to rebuild on the old installer.
This is a very old argument that has been soundly rejected. If you don't actively work on the installer you get no say because that's how OSS works, simply put. And the people who actively work on it said this massive work was needed because "rebuilding on the old installer" was not possible. Why? Go read the archives, they discussed in exhausting detail why it needed a rewrite, what it couldn't do that they felt it should be able to do. And why it wasn't just a matter of fixing some parts of it. At the core it wasn't easily customizable and had no concept of basic things that enable teaching it about new storage technologies like Btrfs, LVM thinp, and bcache. You're asking for an old car to be turned into a hybrid after market. Well that's more work and more money with worse results than just buying a new car that is already a hybrid.
However, the new installer is far too removed in process from the old one, and the least that could be done could be to include a manual and an FAQ (question bank) with the installer.
Have you read any of the Fedora 18, 19 or 20 documentation on the new installer? If it was inadequate or confused you did you file bugs against Fedora documentation?
This could be searchable (and included with the installer, because, again note that the machines functionality is not that high at the time of the install). Of course, this takes effort and time which volunteer developers may not have or wish to put into.
There is a Help button, it's the fifth button, in Manual Partitioning that contains some minimal documentation, that actually is pretty useful in understanding the paradigm shift involved. If you accept that information, then the manual partitioner makes a lot more sense. If that text is confusing or inadequate, file a bug against anaconda with suggested improvements, or even questions that you don't have answers to that you think this help section should answer.
It really can't be exhaustive, there isn't enough space for it, although in a Live Desktop install you can have the installer and Firefox running at the same time.
Btw, the new installer is uglier in terms of looks than the old, too, but that is a matter of personal like!
It's an unqualified statement that doesn't at all convey a path to fixing a problem.
My rambling is all to say that while I eventually got my stuff the way I wanted, I am not sure if it should have taken all that while. I had installs happening in 3-4 minutes with the old installer (now the time taken is exponentially higher). So, while I appreciate all the hard work done by the developers, I can not but wonder if their hard work has been in vain (to an extent) in the sense that many of us old and loyal F users are unable to fully appreciate all their contributions in this regard.
This is a nascent UX bug description. It just needs to be made much more specific about what your expectations are, at a particular step, and how the installer betrayed your expectation, or confused you. If you can't articulate this with some better detail in a bug report it really has much less likelihood of getting better.
One area I know that's still considered weak is the installer's handling of existing installations. It doesn't make it easy to e.g. wipe a prior Fedora or Linux installation while keeping its /home for use with the new install. The expectation was that most users would do this with fedup upgrades rather than cleanly installing a new OS and keeping home.
Chris Murphy
On Sat, 2014-03-22 at 16:40 +0000, Liam Proven wrote:
As I have said previously, I have /never/ successfully installed Fedora on actual hardware since v1.0 shipped in, what was it, 2003? I have installed Haiku, Aros, FreeBSD, PC BSD, dozens of Linux distros, Windows 2 through 8, SCO Xenix, SCO Unix, OpenSolaris, OpenVMS, FreeDOS, DR-DOS, MS-DOS, PC-DOS, OS/2 1 through eComStation 2, MacOS 6 through OS X 10.9. I am *not* a newbie and I am *not* an inexperienced inexpert fumbler.
I know you've said this before but I still find it astonishing. I've used Fedora since FC1 and have never failed to install it or update it on real hardware (and not always the same real hardware). Once or twice I've had issues with the layout (in fact I had some with F20 when it decided I wanted a BTRFS subvolume spread over two disks) but they've been the exception. I do agree that the installer needs to be clearer than it is, though I find F20 better than F19.
I'm not questioning your experience, I just find it remarkably different from my own.
poc
On Mar 22, 2014, at 7:28 AM, Timothy Murphy gayleard@eircom.net wrote:
Joe Zeff wrote:
Fedora, OTOH, is a much more geeky distro designed as a test bed for new ideas, programs and technologies that's not for people who don't like to tinker with things or who aren't willing to accept that not everything in their distro is really ready for prime time.
I'm afraid this "bleeding-edge" metaphor is often used as an excuse for missing or poor documentation. Official Fedora documentation is excessively verbose, and not good at answering - or even considering - issues and problems that users are likely to face.
Verbosity is a common problem in technical writing. It's easy for experienced, knowledgeable technical users to just do brain dumps. It's hard for them to adopt prose. It's probably harder still for doc team leaders of a volunteer system to be good editors which means often brutal conformance to the prose standard. It can mean simply not using a LOT of material, which then limits the type and number of documentation contributors. It's not an easy problem to solve.
As published author who was insanely verbose (see these emails, and imagine them being 10x longer), I learned that establishing prose and scope is critical. And you can't take anything personally, or expect to get any feedback or justification on your writing style. That's just too much coddling. So you'd only get into documentation as a totally selfless thankless act because that's the core job description.
Getting people who are good editors, and writers who are totally content with maybe 90% of their content being completely rewritten or tossed, is a tough combination to find in a volunteer project I think.
A major barrier to getting me to contribute to docs is that I'm totally unfamiliar with the publishing tools used and have zero interest in learning them. So I don't know how that gets worked around, or if it's just one of those filters like an LSAT or MCAT. It is possible to effectively contribute by filing bugs against documentation, and I do that. So I suggest filing bugs if you come across something that's really wordy and just not conveying what needs to be conveyed.
There is - or should be - a big difference between a Reference Manual and a User Manual.
As an example (at random) I'd take using a USB stick for installation. Does anybody nowadays actually burn CDs or DVDs? Yet finding instructions on the use of USB sticks (and external USB drives) in the official installation manual involves jumping around from chapter to chapter.
I suggest you bring your concerns to the Docs team email list, if for no other reason than they understand people do care about this sort of thing, and what the pain points are. It's no different than any other kind of criticism of software, or a bug report. In fact most of my doc contributions are in the form of filing bugs because I'm familiar with that entry method and I'm unfamiliar with the tools they use.
Chris Murphy
Matthew Miller mattdm@fedoraproject.org writes:
On Fri, Mar 21, 2014 at 10:55:56AM -0700, Rick Stevens wrote:
Unlike others, however, I find the new system logging and analysis tools cumbersome and painful to use. Having a program send an email to me if it encounters issues is FAR superior to me having to plow through the logs to see if it ran correctly or not.
We definitely need better alerting and monitoring, mail-based or otherwise. Help wanted. :)
Emails are just fine for this. I don`t want or need some place else I need to look at, and I don`t want to be bothered by notifications that pop up somewhere.
Other than that, yes, some notification thing that actually works would be nice for instances when you need it. I had to write my own --- it`s using libsx, for which unfortunately no Fedora package exists ... It should be easy to make it suitable for things otherwise sent by email.
It would seem like effort wasted on replacing something that works great. How about starting an MTA only on demand? Debian had/has that approach.
BTW, what`s the Fedora way of starting eximstats? I haven`t really looked into it yet, it would have to be coordinated with logrotate. I want that report --- by email, of course :)
FWIW:
// sxnotify.c // This software is licensed under the GPL. // Author: lee@yun.yagibdah.de, 2013-07-21 // // compile with something like: // gcc -lsx -lXpm -lXaw -lXt -lX11 -O2 -Wall -fomit-frame-pointer -finline-functions -march=native bwstat.c -o bwstat // // gcc -lsx -lXpm -lXaw -lXt -lX11 -march=native -O3 -finline-functions -ffast-math -fno-math-errno -funsafe-math-optimizations -ffinite-math-only -fno-signed-zeros -fsingle-precision-constant -fcx-limited-range -funroll-loops -ftracer -fvariable-expansion-in-unroller -freorder-blocks-and-partition -flto=4 -fprofile-generate // That should compile without warnings. Now run: //
#include <stdio.h> #include <unistd.h> #include <stdlib.h>
#include <libsx.h>
void xx(void *data) { exit(0); }
int main(int argc, char *argv[] ) {
if( !OpenDisplay(argc, argv) ) { puts("cannot open display"); exit(1); }
if(argc != 2) { puts("usage: sxnotify message"); exit(1); }
MakeLabel(argv[1]); ShowDisplay(); int foo; AddTimeOut(5000, xx, (void *)(&foo)); MainLoop(); exit(0); }
I took libsx from the source package Debian has. It doesn`t compile all the examples, but you can get the lib itself. I wish there was a Fedora package for it ...
Instead of making a label, you can make an editor, which would be better for long messages --- but then, I have emacs running anyway and would just start an emacsclient. So you`d be looking for something that perhaps queues the messages and waits until something to display them is running.
So why not use emails? Like I said, without a replacement for an MTA, you do not have a functional system ...
Liam Proven lproven@gmail.com writes:
On 21 March 2014 23:25, Chris Murphy lists@colorremedies.com wrote:
Ahh, so when hyperbole simply isn't going far enough we actually have to descend into the obviously ridiculous, as in, worthy of ridicule. My house plant can do a Fedora install with this installer. That you keep failing to get any kind of successful installation is a bit amusing. Maybe you need more water?
Firstly, your mocking hectoring tone is very unhelpful, annoying and is not a productive way to engage.
+1
The F20 installer was completely unable to understand it and allow me to install a complete system. Assigned some 250GiB of space, it said that it needed 6.5GB and there wasn't enough room.
Yeah I had the same thing coming up with the F19 installer. There was plenty of room and it said there isn`t and refused to use the partitions that were there. And try to make it use existing partitions ... "disaster" is a good word to describe it.
In trying to install, it erased one of the spare-root partitions and was unable to recreate it in the available empty space.
That`s why I`d have to unplug all disks except the system disks I`d install on --- which is a PITA. And it`s not even possible when you have a laptop and are forced to install on the same disks the data is on.
It *is* broken and it *is* unusable. "Well it works for me" is *not* an adequate reply.
+1
Let`s say you install to a software RAID-1 --- which is minimum requirement for anything to put data on --- made from two disks, with encrypted partitions (as usual /, /usr, /home, /tmp, /var, /usr/local, and a swap partition). You want to have these partitions in a particular order on the disks, i. e. swap at the beginning because chances are it`s faster, then /usr, /var, /tmp, /usr/local and /home, in that order.
That`s nothing complicated, either, and I don`t think that`s possible with Fedoras installer. Or is it? And if it is, how long does it take to do the partitioning?
Joe Zeff joe@zeff.us writes:
to say, designed for "Windows refugees." Fedora, OTOH, is a much more geeky distro designed as a test bed for new ideas, programs and technologies that's not for people who don't like to tinker with things or who aren't willing to accept that not everything in their distro is really ready for prime time.
Hm, what do you really need to tinker with in Fedora (when you let gnome aside which I find now even more unusable than it used to be, and when you don`t mention the installer (with which you can`t tinker anyway))? It`s just working fine (if you get it installed).
And people who don`t want to tinker might just tell the installer to do whatever it wants, and that probably works, too: Lots of people care about their existing data by pretending that it doesn`t exist and/or not knowing that it does or where it does exist, and if in doubt, they`ll say it`s not worth keeping anyway.
Matthew Miller mattdm@fedoraproject.org writes:
On Fri, Mar 21, 2014 at 04:19:05PM +0100, lee wrote:
Because really, even though many people think the base OS is now boring, it’s far from done, and there is a lot of innovation going on at that level as well.
What do you consider as "base OS"?
It's somewhat nebulous, but, as a general working definition, the system stuff below the applications layer. (Not in the OSI sense.)
Are there significant differences in that with different Linux distributions?
And on top of that, what is the Fedora-way of replacing gnome --- which I find totally useless --- with fvwm, which perfectly does what I want?
It sounds like you want to do a minimal install and then add up from that.
Yes, that`s what I always did with Debian.
I think you will benefit from this effort in that the minimal install will be better defined and curated.
That would be nice --- I wouldn`t even have thought that there is one if I hadn`t read on this list that there is one, somewhere. I still don`t know how I would start with a minimal install, though.
When you get the installer and boot it, you get a working system. That`s a good way to go because otherwise you need a second computer around when installing to look up things. But where is the minimal install, and what when you don`t get a GUI?
It`s only one example, and you can figure out how to do it. But the point with this is that Fedora lacks flexibility. You get what you get and then have to go through a lengthy process of getting rid of stuff and of somehow getting to work what you need, like fvwm.
Sure, I would agree that this isn't a strong suit, particularly with the all-or-nothing way RPM dependencies currently work.
Yes, that is one of the things I don`t like. It forces you to install stuff you never need.
But on the other hand, I'm not sold on it being a huge problem. If you know what you are doing, it's not that big of a burden, and I'm pretty sure that the intersection of people who want this and people for whom it is easy with Fedora as it stands is quite large.
Well, look around at what ppl say about different distributions like you would in order to decide which one to use. What you find out is primarily what "desktop environment" and, in a side note, what kind of package management they are using.
I still have no use for what`s called a "desktop environment". They slow down my computer with all kinds of stuff I don`t need and, more importantly, get majorly into my way. Try out several, and you find that when you get set up in one, you have to start over to get what you need when you try out another one. Not even the keyboard is working right because your ~/.Xmodmap is being ignored.
Yet what "desktop environment" is the default seems to have become the most important feature of a Linux distribution. Otherwise it might be mentioned in a side note, if at all, and important things would be pointed out instead.
Having choices is one of the most important things. You still have them with Fedora --- otherwise I wouldn`t be using it. That kinda makes it relevant how easy or difficult it is to get what you want.
The point is that you need to know what you`re doing, i. e. how to get the choices. Why not make it easy for people to choose?
On a side note: The installer sucks, just try to do one of the most basic and important things with it: Partitioning.
Saying something "sucks" isn't very helpful. Not only is it needlessly negative, it is intangible. Name a real problem and we can talk about it.
There have been a few threads about it on this list. The major problem is partitioning.
The last time I used the installer, it was the one F19 comes with. It was impossible to get the partitioning I wanted, so I had to partition otherwise. Then it was nearly impossible to make the installer actually use the partitions the way I wanted. And since the buttons the installer uses are weird and misleading, you never really know what you`re doing. I had to try over and over again to figure out how to somehow make it use the existing partitions. If I had used it on a computer that had data on the disks I wanted to keep, I`d have had to physically unplug them to make sure the data doesn`t get lost --- and that isn`t always possible.
So some simple partitioning that would be done with the Debian installer within ten minutes took three hours. The Fedora installer does what it want`s, not what the user wants, and it leaves the user in the dark about what is actually going to happen.
On top of that, it doesn`t offer any choices. You cannot pick any package at all.
Take a look at the Debian installer. I never used their graphical one --- an installer shouldn`t require a GUI; a GUI is just annoying for that. It`s been a while since I used it, but it`s worlds ahead of the installer Fedora has. It`s not perfect, either, though.
On the specific you do give, I'm pretty confident in saying that you're actually wrong.
Unless the installer majorly changed from F19 to F20, I`m not wrong.
Storage is hard, and the new anaconda contains the most sophisticated and powerful GUI partitioning tool ever made.
Seriously? And like I said, I don`t like GUI installers at all.
A GUI may work or not. I`ve seen them not working too much to rely on them, especially not when installing an OS.
It's just a really difficult area to get perfect.
It`s very simple, actually. You only need to give the user the possibility to use and to create partitions the way they want. You also need to be perfectly clear --- and this shouldn`t even need to be mentioned --- about what will happen to the partitions.
Expect this to continue to improve with each Fedora release as the design is refined and bugs are shaken out.
Hopefully so ... I found the installer of F17 much better in that point, even though it didn`t let me use a separate partition for /usr.
And when you managed that, you can`t start with a minimal install and install just what you need.
You can start with a pretty basic install in the GUI.
How? You get an icon you click on and from there on, you nowhere see that choice.
If you need more than that, you really should be looking at kickstart, and maybe even producing your own spin.
Why does it need to be a spin? Why can`t we have choices?
I was at a large cloud conference a while ago, and almost nobody was using Fedora, and so I asked people why they chose the distribution they are building their stuff on, and why they didn’t choose Fedora. Almost universally, the response wasn’t “What I am using is great!” — it was “Oh, I don’t care. I just picked this, and that’s what I’m using and it’s fine.”
I`m not one of these people. Thinking like that, they don`t need a Linux distribution; they can as well use Windoze or Macos.
Yet these people were absolutely running Linux. Just not ours.
Hm. Maybe that`s how people are.
Think of other things they use: How many times do they pick something because it`s great? It involves more work to do that because you need to find out what is available, what the differences are and what makes something great for you in particular. So they don`t bother and use something which is fine.
Look at http://fedoraproject.org/ --- it says this and that and doesn`t say or show that Fedora is great or awesome and why.
One thing Fedora shines with (so far) is reliability, and reliability is one of the requirements I have. I have been using Debian for almost twenty years until they messed up badly with their brokenarch. Doing that put Debian out of the question once and for all because they failed that requirement miserably beyond believe. Please do not make the same mistake with Fedora. Switching to another distribution is a painful process.
I don't know what that specific problem was with Debian, but we are certainly working at increasing reliability, particularly through automated testing.
It wasn`t something slipping through quality control, it was their decision. They removed crucial libraries that were working fine for years and left people stranded. There was no fix or replacement for it, no documentation to figure something out. I got a few suggestions to install other packages instead, and none of them worked. The libraries were just gone, and you suddenly couldn`t run the software you used to run anymore.
Letting users down like that is a big mistake. There was no choice but to switch, not only because the software didn`t run anymore: It clearly shows that the makers of the distribution don`t care, and once they start doing things like that, it is foreseeable that there will be more trouble in the future.
There can always be a bug somewhere that has gone unnoticed. They are likely to be found and to get fixed --- usually no big deal. Some are or can not be fixed, or it takes long before they get fixed, and you may have to find another solution. Unless there are too many bugs, it`s not a problem for a distribution. Fedora already does a good job with it.
But when the makers of a distribution turn their backs on the users like Debian did, you have to replace that distribution with something else.
But I find the reliability I`ve seen so far with Fedora very exciting and I am very thankful for it. It is a great accomplishment of Fedora.
As I started with: Fedora is awesome.:)
:)
Add more choices, and it will be more awesome :)
Also: the base OS has developed to the point where it has become uninteresting.
Perhaps not: It is possible that the expectations have changed. Ten years ago people were probably much more willing to accept that their soft- or hardware doesn`t work and that their computer crashes or freezes every now and then than they are willing to accept this nowadays. Some people still don`t care.
Actually, the change is the opposite, at least in the context I am talking about. Check this out: https://www.usenix.org/conference/lisa13/how-netflix-embraces-failure-improv...
Hm, IIUC, the idea is to figure out when/how/why something fails in order to make it so that it doesn`t fail anymore. That`s not really new?
The people you have been asking why they aren`t using Fedora probably were not people using Linux for the purpose of testing Linux distributions. They were probably people who found something that works.
For Linux distributions, "great" and "awesome" are the default. People who are looking for something that works pick a distribution and it`s fine because it meets their expectations because it works. That makes it appear as if it has become irrelevant which particular distribution they picked.
It`s a recipe for failure because not all distributions work evenly well. People are not aware of that until they experience the failure.
Ten years ago, what to pick may have appeared more important because people experienced and thus expected more failures. That made them think more about what to pick, which made it appear more relevant which distribution that would be.
I didn`t pick Fedora because I expected greater reliability, but for other reasons. Now that I`m experiencing this greater reliability, it is a very strong point that speaks for Fedora. I used to put Debian on servers. Now I`d put RHEL or Centos instead just because of that and because Fedoras pace can be scaringly fast. And if I had known that Fedora works so much better than Debian, I might have switched long ago.
The problem is that you don`t find out these things when you want to pick a distribution. You just have to pick one.
and to make that easy, I`d like to make a package". Then you try to find out how to do that and that`s where it ends: It`s just too difficult. Instead, you put your software on github.
Absolutely. We need to make that easier.
How about some documentation and a mailing list for it? Somehow, the people making a distribution --- and those are the ones who make packages, I guess --- seem awfully far away from the users.
That decision came across as "removing an MTA from the default install". I don`t know if you`re saying that something is now replacing sendmail or that there is no MTA when you do a default install. If it`s the latter, no MTA at all, then it was a very bad decision. A system without MTA is not functional.
Sure it is.
It is not without a replacement. There`s lots of software sending you stuff by email.
Again, hyperbole doesn't really help. In fact, in many situations, an MTA doesn't do you any good, as your network won't allow it to do anything useful remotely (true now at most big companies and on most big ISPs)
It doesn`t have to do anything not local to be required.
and local delivery goes into the black hole of root's mailbox unless you configure it.
Mail disappearing without notice is unacceptable. If Fedora has such a configuration with a default install, then that needs to be fixed.
And if you're going to configure one, installing one isn't a significant extra step. Plus, since there are multiple different MTAs, this is back to your choice of fvwm vs. gnome -- if we have a default you don't want, you're going have to remove it to put in your choice.
It`s perfectly fine not to install an MTA when there is a replacement that handles the local mail. Without a replacement, you don`t have a working system until you install an MTA.
So that’s part of what Fedora.next is: to look at our mission and decide what more we need to do to make it happen.
Let me simply ask what "community" is supposed to mean in the mission statement (which I can`t quote here because it`s an image).
https://fedoraproject.org/wiki/Overview#Our_Mission
"The Fedora Project's mission is to lead the advancement of free and open source software and content as a collaborative community."
I am asking this because nowadays, everything is a "community", to the point where that word doesn`t mean anything anymore. What, who and where is this "community" in this case, who are the members of it and how does one become a member?
"Collaborative community". That means it is all the people who work together on the project. You can become a member by saying you are and doing something.
So I was right to think that being collaborative is substantial and constituting for "community" in this case. I forgot about "leading", though. And you could say that there are some people who want to work together to lead the advancement of FOSS.
Why not just say that? Things can be said simply when you understand them, and when something cannot be said simply, you have an indication that it cannot be understood.
Anyway, thinking of it, I`m finding the aspect of "leadership of advancement" quite remarkable. How is that to be accomplished?
That's all there is to it. (Although, practically speaking, you will also want a Fedora account if you don't have one.) See https://fedoraproject.org/wiki/Join for a lot more starting material.
Thank you, I`ll look into it.
Listen to the users, if you want to make a better distribution.
Sure. Continue speaking up -- we are listening.
Good :)
Maybe we could do a better job by letting people put their packages in
Assuming that they can make packages ...
Excellent point. Not requiring packages is the next level.
Hm, I thought you`d say something like that. Someone still needs to make them, though.
I think I`d like to see an article with questions like this which lays out what answers to these questions are currently in place. There probably aren`t very many people who know the current answers. Such an article would also need to explain for each question why it is necessary to ask it now.
I'll think about it. :)
I`ll read it when you write one :)
On Sat, Mar 22, 2014 at 05:38:48PM +0100, lee wrote:
What do you consider as "base OS"?
It's somewhat nebulous, but, as a general working definition, the system stuff below the applications layer. (Not in the OSI sense.)
Are there significant differences in that with different Linux distributions?
It depends on what you mean by "significant". From some points of view, it's very, very different -- different init systems (although that seems to be slowly converging on systmed), different packaging, different logging conventions. Or maybe you care that there are different library and even kernel versions. But from other points of view, that's all implementation details.
And on top of that, what is the Fedora-way of replacing gnome --- which I find totally useless --- with fvwm, which perfectly does what I want?
It sounds like you want to do a minimal install and then add up from that.
Yes, that`s what I always did with Debian.
You can do that on Fedora, although minimal isn't quite as small.
I think you will benefit from this effort in that the minimal install will be better defined and curated.
That would be nice --- I wouldn`t even have thought that there is one if I hadn`t read on this list that there is one, somewhere. I still don`t know how I would start with a minimal install, though.
In the GUI installer, under software selection, scroll to the bottom and pick "Minimal Install".
In kickstart, close-to-minimal is the default, and you can use "%packages --nocore" if you're really serious.
When you get the installer and boot it, you get a working system. That`s a good way to go because otherwise you need a second computer around when installing to look up things. But where is the minimal install, and what when you don`t get a GUI?
The above will start you without a GUI.
[snip]
Take a look at the Debian installer. I never used their graphical one --- an installer shouldn`t require a GUI; a GUI is just annoying for that. It`s been a while since I used it, but it`s worlds ahead of the installer Fedora has. It`s not perfect, either, though.
You can use the installer in text mode. You can also use it in completely scripted mode.
On the specific you do give, I'm pretty confident in saying that you're actually wrong.
Unless the installer majorly changed from F19 to F20, I`m not wrong.
Storage is hard, and the new anaconda contains the most sophisticated and powerful GUI partitioning tool ever made.
Seriously? And like I said, I don`t like GUI installers at all.
Seriously. And since you don't want a GUI installer, you probably should preconfigure your disk with whatever tools you want and then install onto it with kickstart.
(more on the rest in a separate reply)
On 03/23/14 00:42, Liam Proven wrote:
On 22 March 2014 05:00, Ed Greshko ed.greshko@greshko.com wrote:
So, I don't know what may be different in my configuration (maybe the windows swap?) that allowed me to compete the exercise. But surely, it isn't totally broken for everyone.
I am not for a moment saying that it /is/ broken for everyone. Just for me, in a moderately sophisticated but not massively difficult multiboot scenario with 3 other distros and Windows.
Make sure you file a bugzilla. Talking about it on this list will not result in any action. Of course, you'll have to be able to consistently repeat the problem and be willing to document it well.
FWIW, for problems such as this one I just happen to have spare disks that aren't bad. They were just replaced for ones of larger capacity. I understand you may not be in the same position.
Ed Greshko wrote:
Does anybody nowadays actually burn CDs or DVDs?
Yes..... The BIOS on many of the machines I encounter don't support booting from USB.
How old are they?
Assuming they are very old, would it not be more rational for the documentation to assume that the user was at a reasonably new machine, and add a section or appendix to deal with the case of an ancient BIOS?
In my view, if 90% of users are going to use option A, then 90% of the user documentation should deal with that case. Treating all options equally is a recipe for verbosity.
Hi Chris,
On Sat, Mar 22, 2014 at 11:21:56AM -0600, Chris Murphy wrote:
Verbosity is a common problem in technical writing. It's easy for experienced, knowledgeable technical users to just do brain dumps. It's hard for them to adopt prose. It's probably harder still for doc team leaders of a volunteer system to be good editors which means often brutal conformance to the prose standard. It can mean simply not using a LOT of material, which then limits the type and number of documentation contributors. It's not an easy problem to solve.
As published author who was insanely verbose (see these emails, and imagine them being 10x longer), I learned that establishing prose and scope is critical. And you can't take anything personally, or expect to get any feedback or justification on your writing style. That's just too much coddling. So you'd only get into documentation as a totally selfless thankless act because that's the core job description.
Getting people who are good editors, and writers who are totally content with maybe 90% of their content being completely rewritten or tossed, is a tough combination to find in a volunteer project I think.
A major barrier to getting me to contribute to docs is that I'm totally unfamiliar with the publishing tools used and have zero interest in learning them. So I don't know how that gets worked around, or if it's just one of those filters like an LSAT or MCAT. It is possible to effectively contribute by filing bugs against documentation, and I do that. So I suggest filing bugs if you come across something that's really wordy and just not conveying what needs to be conveyed.
Is anyone from the Fedora docs team attending the newly started "Write The Docs" conference[1]? I think this is not as tough a problem as many make out to be. A few examples would be Archlinux and Gentoo; despite being completely volunteer driven distros (no backing by a company like Red Hat) they do pretty well. I personally sometimes contribute to documentation efforts for an Emacs major mode, Org mode[2]. It is definitely not easy, but it is quite possible to have reasonably comprehensive documentation.
I think the problem with Fedora is more focus on user manual/guide like documentation rather than references. We should focus on having a comprehensive reference-style docs first before jumping towards guides/manuals. In fact, in my experience if reliable reference-style docs exist (say written by experts: devs interested in docs, knowledgeable volunteers, ..), users are perfectly placed to create the manual/guide style docs.
Just a few thoughts.
Footnotes:
[1] http://conf.writethedocs.org/eu/2014/ I'll be there, if anyone is interested. [2] http://orgmode.org/
On Mar 22, 2014, at 10:40 AM, Liam Proven lproven@gmail.com wrote:
On 22 March 2014 03:54, Chris Murphy lists@colorremedies.com wrote:
Hmm. Now I believe you were just about to cite a bugzilla ID describing the above behavior?
No. Why should I? Total failure to install the OS leaves me unable to use its bug-tracking tools, if any.
Because you apparently expect it to work, and yet it doesn't work as expected (or otherwise)?
As I have said previously, I have /never/ successfully installed Fedora on actual hardware since v1.0 shipped in, what was it, 2003?
Ten years of failure installing Fedora. You should make your own t-shirt.
In those ten years, you're still unfamiliar with the installer's bug reporting mechanism? In lieu of the fact you have a 0% install success rate installing Fedora on baremetal, aren't you suspicious? Curious?
Have you inquired with anyone about how to get more information from the installer to possibly find out why you've had a decade of install failures, on just Fedora? Are you unfamiliar with the installer's rather substantial logging feature, always enabled, with easily recovered logs? Have you looked at them, or had anyone else look at them? What was the cause of the failure?
I have to tell you that in my experience of approaching a hundred installers, it is about the least capable of any C21 OS I have ever seen. I think it might beat eComStation but nothing else.
One hundred installers, OK you beat me.
You have Apple hardware, which I have, and I've done more successful Fedora installs on those three models in the past year alone than the number of installers you've used. Seems conspicuous.
CentOS, yes, although I've seen it fail too. Fedora, nope.
You realize that CentOS and Fedora use the same installer, right?
And you have a bug for this?
No, what I have is material for a really bad product review.
What you've presented here so far is merely conjecture. Material would be bugs, logs, maybe a detailed spec of the computer involved and video of the failed install process. You've cited no error messages, indicating the exact nature of the install failure. You haven't presented anywhere near enough information for someone to reproduce your results.
I can hardly imagine a competent, ethical journalist nor his editor accepting such lack of evidence for a negative product review. Such a negative product review would be egregiously discourteous to the development team, and I think the project as a whole. I even think it would be unprofessional, and by that I mean a professional uses proper tools for the job. In the course of testing a product, a professional uses normally available debugging tools to better understand the nature of the failures experienced at least in some basic level to ensure the problem isn't user error or related to unsupported hardware, or myriad other explanations.
This is understandably confusing if you're really familiar with storage stack creation. But most people aren't.
I hope that I have demonstrated that I am not unfamiliar.
Oh, you have demonstrated something, that much is certain.
But I assure you that the vast majority of the world's installers would have poo poo'd you much sooner.
And again, no, you're wrong. Sorry, but you are and I can, if you wish, show you.
By all means produce installer logs for your failures, ideally as attachments to bug reports.
Chris Murphy
On 03/23/14 02:29, Timothy Murphy wrote:
Ed Greshko wrote:
Does anybody nowadays actually burn CDs or DVDs? Yes..... The BIOS on many of the machines I encounter don't support booting from USB.
How old are they?
Probably 2~3 years old. They have either i5 or i7 CPU's. Some are HP branded, but the motherboard is MSI.
Assuming they are very old, would it not be more rational for the documentation to assume that the user was at a reasonably new machine, and add a section or appendix to deal with the case of an ancient BIOS?
The "easy" way to deal with these "not so old and perfectly good" systems is to use a DVD/CD that have been burned.
In my view, if 90% of users are going to use option A, then 90% of the user documentation should deal with that case. Treating all options equally is a recipe for verbosity.
Well, the question you asked was "Does anybody nowadays actually burn CDs or DVDs?". And is exactly and only what I'm addressing.
Yes, documentation could be better. And, you're perfectly able to assist in correcting that deficiency.
Hi,
You do have a legitimate point about enough users not putting in BZ requests, however, I have grown very frustrated with the BZ process (for a few other packages). Lazy maintainers (and let me throw away my restraint to call them for whatever they are here) are using the flimsiest of reasons to deny legitimate BZ requests. Case in point: zathura which seriously needs an upgrade from before the early days of F19 and every request being denied on flimsy pretexts. Same goes for many other packages (LXDE spin's inability to handle ssh-agent since the days of F14, etc, where I even tracked and submitted where the bug was). Yes, these are all volunteer maintainers but still, it would help if there was consistency in how these requests are responded to. (Many of my experiences have been very good in the past.) Currently, I don't even know if it is worth it for me to put in the effort to put together a BZ request because I don't know if the maintainer will even bother applying his mind to the request.
I'd like to see MB rounded to integers and GB rounded to one decimal place except in the lower left Available Space and Total Space fields, which should also be integers.
What about if I want to fill the remaining space with the desired partition? I believe that the answer is that you leave the space allocation blank (as someone in this mailing list told me a few months ago), but I would never have figured this out -- why did that option from the old installer get incorporated into the new, or why was not the fact that it has to be kept blank not mentioned during the install partitioning process options?
Months later, I have figured out that there are buttons to set the time according to a desired network provided. It is all hidden, and I accidentally found it by looking and wondering why this option should not have been included in the install options.
I don't know what this refers to. Buttons to set time per desired network?
OK, what i was referring to was the ability to set the time as per chosen network time provider (and to provide a network time provider). This was not easy to figure out at all.
Btw, the new installer is uglier in terms of looks than the old, too, but that is a matter of personal like!
It's an unqualified statement that doesn't at all convey a path to fixing a problem.
Agreed, but this is the least of my concerns. I believe that the installer is ugly, but it is a matter of personal likes and not something that I care to mention.
One area I know that's still considered weak is the installer's handling of existing installations. It doesn't make it easy to e.g. wipe a prior Fedora or Linux installation while keeping its /home for use with the new install. The expectation was that most users would do this with fedup upgrades rather than cleanly installing a new OS and keeping home.
Agreed again. This, however, brings in more problems and gripes for when a new F has to be installed for the first time on a new system (typically, once every few years when a machine is bought), and the shock of discovering something that is so suddenly out of whack is then substantially more.
Best wishes, Ranjan
____________________________________________________________ FREE ONLINE PHOTOSHARING - Share your photos online with your friends and family! Visit http://www.inbox.com/photosharing to find out more!
"Powell, Michael" Michael_Powell@mentor.com writes:
Absolute ease of use is what they're going for and the easiest way to achieve that is to reduce the amount of information a user has to process.
Then they need to turn around at once! Leaving users in the dark about what`s going on makes things more difficult for them. Taking away choices limits the use of the software to the point where it eventually becomes unusable.
Good design, clarity, good documentation and letting the user know what choices there are are required when you want to achieve ease of use.
Ed Greshko ed.greshko@greshko.com writes:
- Booted into Ubuntu and used parted to create dummy unused partitions.
- Then did the F20 installation and selected dummy partitions for reclamation and the installation completed.
Creating the partitions otherwise does make a difference. With the F19 installer, it was impossible to get the partitioning I needed. Once I had them created otherwise, it still was a PITA to get the installer to use them the way I wanted.
Matthew Miller mattdm@fedoraproject.org writes:
On Sat, Mar 22, 2014 at 05:38:48PM +0100, lee wrote:
What do you consider as "base OS"?
It's somewhat nebulous, but, as a general working definition, the system stuff below the applications layer. (Not in the OSI sense.)
Are there significant differences in that with different Linux distributions?
It depends on what you mean by "significant". From some points of view, it's very, very different -- different init systems (although that seems to be slowly converging on systmed), different packaging, different logging conventions. Or maybe you care that there are different library and even kernel versions. But from other points of view, that's all implementation details.
I`d say that there aren`t any significant differences then: It doesn`t matter for a particular software which package management is used, whether there are different logging conventions or how the software is started. When your MTA or library or kernel has a bug, it has the bug regardless, unless the makers of the distribution perhaps modify or fix things so that they have a distribution-specific version. Different versions do matter, though that averages out because any version can have a problem: One distribution may happen to skip a particular version which is buggy while another distribution has it. Five versions later it can be the other way round.
The difference is for the user who likes one package management system better than another, or prefers one init system over the other.
It leads to an interesting question, though: How come that one distribution works better than another despite there is no significant difference in the software they are using?
And on top of that, what is the Fedora-way of replacing gnome --- which I find totally useless --- with fvwm, which perfectly does what I want?
It sounds like you want to do a minimal install and then add up from that.
Yes, that`s what I always did with Debian.
You can do that on Fedora, although minimal isn't quite as small.
Hm, I never noticed a choice like that ... "Minimal" isn`t about "small" in this case, it`s about not installing what I don`t need.
BTW: You were saying an MTA has been removed from the default installation because it cannot do anything useful, especially not in an office environment, unless appropriately configured, and that when you configure it, it doesn`t make a difference when you need to install it first.
The same goes for pulseaudio. It doesn`t do anything useful, especially not in an office environment where soundcards are usually not used, and not at all when you don`t want to use a soundcard. In case it could actually do something useful for you, you have to configure it to do so, and in that case, it doesn`t make a difference when you need to install it first. So why is it not only installed by default but impossible to get rid of it without removing gdm and some other packages?
I don`t need pulseaudio. Alsa handles things anyway and does it just fine without. All that pulseaudio does is that it adds an obsolete layer and wastes CPU time in doing so. I might use gdm in case I want to start a gnome session, though.
And when I log in a second user, last time I checked the second user doesn`t have any sound. I can`t get it to do DRC, either. So why is pulseaudio installed? Unlike an MTA, it serves no purpose at all.
I think you will benefit from this effort in that the minimal install will be better defined and curated.
That would be nice --- I wouldn`t even have thought that there is one if I hadn`t read on this list that there is one, somewhere. I still don`t know how I would start with a minimal install, though.
In the GUI installer, under software selection, scroll to the bottom and pick "Minimal Install".
It should ask me what I want to install ... Debian does it right, their installer has many steps, and you can even go back and re-do one, or do some in a different order. One of the steps is to pick what you want to install.
In kickstart, close-to-minimal is the default, and you can use "%packages --nocore" if you're really serious.
I don`t even know what kickstart is ...
You can use the installer in text mode. You can also use it in completely scripted mode.
That`s cool :) How do you do that?
On the specific you do give, I'm pretty confident in saying that you're actually wrong.
Unless the installer majorly changed from F19 to F20, I`m not wrong.
Storage is hard, and the new anaconda contains the most sophisticated and powerful GUI partitioning tool ever made.
Seriously? And like I said, I don`t like GUI installers at all.
Seriously.
Then it must have been re-designed from scratch since F19.
And since you don't want a GUI installer, you probably should preconfigure your disk with whatever tools you want and then install onto it with kickstart.
That depends on what you consider should be part of an installer and what not. I think it should be possible to get the partitioning you want without using additional tools beside the installer. That doesn`t mean that the tools have to be built-in. The installer could give you a choice to use cfdisk or parted or something built-in. It doesn`t even need something built-in to do the partitioning, though. It would suffice to tell it what to do with which partition.
Powell, Michael sent:
The system name or hostname is important to networking; so, I can see why it's under networking, but I believe your frustration is more related to the lack of guidance and quality than anything else.
Tim:
Well, actually, for a lot of people, the system name is simply what they want to call the computer. The computer may not even be on a network, at all. There's certainly cause for having a process of naming the computer.
for having a *non-network* process of naming it...
Which brings to mind an annoying situation that crops up from time to time: Booting up a computer that's usually connected to a network, but sometimes not. For those "not" times, you may find your computer has been renamed "localhost," which is annoying at best, and problematic at worst (if you have some things that used the old hostname, and expect it to still be in use).
Likewise for connecting that same computer to another network, which gives it a different hostname.
On Sat, 22 Mar 2014 18:29:02 +0000 Timothy Murphy gayleard@eircom.net wrote:
Ed Greshko wrote:
Does anybody nowadays actually burn CDs or DVDs?
Yes..... The BIOS on many of the machines I encounter don't support booting from USB.
How old are they?
I have 2013 manufactured boards, I don't use USB booting or install. I still use LiveCD.
___ Regards Frank frankly3d.com
On 03/23/2014 08:56 AM, Frank Murphy wrote:
On Sat, 22 Mar 2014 18:29:02 +0000 Timothy Murphy gayleard@eircom.net wrote:
Ed Greshko wrote:
Does anybody nowadays actually burn CDs or DVDs?
Yes, because unlike USB-sticks, I am archiving them and resort to using them as "ready-to-use/off-the-bookshelf" media, e.g. in cases of "emergencies".
Ralf
Timothy Murphy:
Does anybody nowadays actually burn CDs or DVDs?
Ed Greshko:
Yes..... The BIOS on many of the machines I encounter don't support booting from USB.
Or the BIOS is none-too-great at booting from USB.
Just recently I had to boot from USB to install onto a laptop with no optical drive. I found four different methods of putting a Linux installer onto a USB. One of them involved Windows, so that was out. Two of them were Linux programs for writing installers to the flash drive. One required adding things to a current install that I didn't want to do. The other was less painful, and worked for writing one installer to a USB, but not for another (I tried Fedora and Ubuntu on this laptop). The last was plain dd if=/install.iso of=/def/flashdrive (I'm paraphrasing), and that worked for the one that didn't work the other way.
All of which was more hassle than burning a disc, and was actually slower than burning a disc for one of the methods (I don't recall which, but I tend to think it was the dd method).
So, while you can install from a USB, it's not always the most convenient.
On 21 March 2014 15:19, lee lee@yun.yagibdah.de wrote:
Matthew Miller mattdm@fedoraproject.org writes:
I was at a large cloud conference a while ago, and almost nobody was using Fedora, and so I asked people why they chose the distribution they are building their stuff on, and why they didn’t choose Fedora. Almost universally, the response wasn’t “What I am using is great!” — it was “Oh, I don’t care. I just picked this, and that’s what I’m using and it’s fine.”
I`m not one of these people. Thinking like that, they don`t need a Linux distribution; they can as well use Windoze or Macos.
Sigh. 1. "Windows" not "Windoze" 2. If we're being picky "MacOS" 3. Most Linux distros are *much* more similar to each other than they are to Windows or Mac. I use about three different distros at work and have another two different ones at home (though one of those is a Pi, maybe count as a half). I don't find this particularly confusing. If they get on okay with the desktop on whatever they're using then more power to them.
Even if there really are a lot of interesting things going on, people who are trying to actually do things with the distribution don’t attach much importance to them.
Not paying attention to the soft- and hardware that is the basis of what you`re actually doing is a recipe for failure, and for making things difficult on yourself.
One thing Fedora shines with (so far) is reliability, and reliability is one of the requirements I have. I have been using Debian for almost twenty years until they messed up badly with their brokenarch. Doing that put Debian out of the question once and for all because they failed that requirement miserably beyond believe.
Please do not make the same mistake with Fedora. Switching to another distribution is a painful process.
I'm kind of interested to know what your use case is that makes switching distros more painful than fixing a broken install. The only guess I can make is multiple systems.
In turn, this leads to a shift in the balance between the effort to get software into a distribution and the reward of doing so. It used to be that if you had open source software, and you could convince the distros to get your get your software into a distro, that’s how you knew that you had arrived.
I have always wondered how people manage to create packages, for Debian or Fedora. I looked into it because I would like to provide packages, and I found it requires an insurmountable effort. You start with "I have written this software" and get to "I would like that ppl use it, and to make that easy, I`d like to make a package". Then you try to find out how to do that and that`s where it ends: It`s just too difficult.
Instead, you put your software on github.
If you see no value in packaging I'm surprised you aren't using a roll-your-own distro instead. It sounds like that would be a much better match for your requirements of control over everything on the machine and a minimalist system.
Tim wrote:
Just recently I had to boot from USB to install onto a laptop with no optical drive. I found four different methods of putting a Linux installer onto a USB.
...
All of which was more hassle than burning a disc, and was actually slower than burning a disc for one of the methods (I don't recall which, but I tend to think it was the dd method).
So, while you can install from a USB, it's not always the most convenient.
This is exactly my experience. It _is_ more convenient to install from DVD, but the reason for this is that booting from USB is not well documented.
Incidentally, Ed and others have said that there are many machines which will not boot from USB. But surely there are many more that have no CD/DVD drive? And the proportion of these is surely increasing?
But all this is pure speculation. One of the weaknesses of Fedora, in my view, is the apparent lack of interest in what users actually want or need. For example, what proportion of users choose KDE or Gnome, or some other Desktop Environment? Does anybody know? Suppose it turned out that far more use KDE than Gnome. Would that not suggest that KDE should be made the default?
Again, how many are installing on a machine that is already partitioned? Suppose (my guess) it is something like 90%. Would it not be more rational in that case to assume as default that users would prefer to use that partitioning?
I don't think I've ever seen an official or semi-official web enquiry on issues like this.
On 22 March 2014 16:40, Liam Proven lproven@gmail.com wrote:
On 22 March 2014 03:54, Chris Murphy lists@colorremedies.com wrote:
Are you asserting that people who bitch and whine about the installer are entitled to a monopoly on mocking hectoring tone; and unhelpful, annoying, unproductive engagement?
I don't think any of us should do it, ideally. But perhaps in gentle chiding of someone who could have answered their question with Google in 10sec, then OK.
When someone is unable to even install the whole OS, no. Inappropriate.
It is actually a complex layout.
I beg to differ. It is not /trivial/ but it is not complete.
Most of the world's installers can't deal with what you just described.
Factually incorrect. Windows 7, Ubuntu 13.10, Debian 7, Crunchbang and Elementary OS all had no problems.
On my desktop PC, I have a similar layout with Windows 8, Mac OS X 10.6 and Ubuntu 13.10. Again, no problems at all.
The #1 OS install today is software restore [...]
In my extensive experience of OSes going back to when I entered the business in 1988, following about 6-7y as a hobbyist, this is incorrect. You're describing one OS, principally - modern Windows.
So, I read "#1 OS install today" as the installer for the #1 OS. There were a couple of other possible interpretations, but seeing #2 as the Windows retail/upgrade and #3 as Mac that made the most sense.
I'm not going to disagree about the installer flexibility though, the new installer is strange.
The Fedora 20 installer's default/easy/guided/auto path installs to free space. Yet it has more options and outcomes than the total number of all possible options in both the Windows and OS X installers combined.
That is not my direct personal experience. I can demonstrate what I mean with screenshots and comparative step-by-step walkthroughs.
Hmm. Now I believe you were just about to cite a bugzilla ID describing the above behavior?
No. Why should I? Total failure to install the OS leaves me unable to use its bug-tracking tools, if any.
https://bugzilla.redhat.com/ can be accessed from most browsers. If you have screenshots and step by step walkthroughs that would help improve the installer then please send them there. Having them languishing on your computer and complaining on the users list is not likely to result in changes. Writing review articles that say "didn't work" as some kind of revenge on the developers without giving them any kind of chance to sort it out, even for the future, is just vindictive.
FWIW, I also tried installing on a completely empty standalone 250GB USB hard disk. It failed on that, too - it hung after the process began and never recovered. After about 6 hours, I power-cycled the machine.
As I have said previously, I have /never/ successfully installed Fedora on actual hardware since v1.0 shipped in, what was it, 2003? I have installed Haiku, Aros, FreeBSD, PC BSD, dozens of Linux distros, Windows 2 through 8, SCO Xenix, SCO Unix, OpenSolaris, OpenVMS, FreeDOS, DR-DOS, MS-DOS, PC-DOS, OS/2 1 through eComStation 2, MacOS 6 through OS X 10.9. I am *not* a newbie and I am *not* an inexperienced inexpert fumbler.
I suppose I should say at this point I've installed Fedora successfully a few times over the years without trouble (can't give exact count as upgraded a couple of times instead, but F20 and F19 spins most recently on two different machines). I've never found a version of fedora that failed to install on the hardware I was using. Again, if you're finding it wont install on something that's a bug and it needs to be in bugzilla to have a hope of getting fixed. (Doesn't guarantee it will get fixed, but it's a start.)
I've only got Fedora running in VMs. Even Slackware is easier.
I've done hundreds of hours of installer testing over the last year. It has been really frustrating. This is the most complicated/capable installer I've ever worked with other than maybe the OpenSUSE installer. Out of the gate it offerred too much compared to the time/resources allotted for QA, debugging, and code changes needed.
I have to tell you that in my experience of approaching a hundred installers, it is about the least capable of any C21 OS I have ever seen. I think it might beat eComStation but nothing else.
It sounds like either you've been spectacularly unlucky installing Fedora or have some particular requirement (hardware, install setup) that it is failing to deal with.
On Sun, 23 Mar 2014, Timothy Murphy wrote:
Tim wrote:
[snip] But all this is pure speculation. One of the weaknesses of Fedora, in my view, is the apparent lack of interest in what users actually want or need.
[snip]
I don't think it's a total lack of interest. I think that it's an issue of prioritization. From what I've read, the purpose of Fedora is to use the open source community to examine and debug things for potential inclusion in the Red Hat Enterprise products. Accordingly, it is our "job" to get stuck with stuff we don't like and things that are not quite ready for prime time. We get a "bleeding edge" full-featured, well-supported OS to play with, but it's not bulletproof or necessarily the easiest to use. We provide service at beta-testers and test audiences. My impression is that Red Hat is *very* interested in what we put in bugzilla, and *is* interested in usability and preference issues from us -- but that the debugging part is first, usability second, and preference third in the priority list.
I suspect that the preferences of the paying customers come first, and that's as it should be. There are a number of pretty bullet-proof linux distros around, but they are a bit boring. If I wanted that, and I had a preference for RH style stuff, I'd be running CentOS rather than Fedora.
Using Fedora is like getting a big box of presents every six months sent to me by a rather absent-minded elderly aunt, who can't quite remember my tastes. The good part is that it is a big box full of neat stuff. The bad part is that some of it got broken in the mail and some of it I just don't care for. But I throw the stuff I don't like away and enjoy playing with the stuff I like.
billo
On 03/23/14 20:15, Timothy Murphy wrote:
Incidentally, Ed and others have said that there are many machines which will not boot from USB. But surely there are many more that have no CD/DVD drive? And the proportion of these is surely increasing?
I have no idea about actual numbers and unless one does the research it is pure, as you said, speculation.
I can, however, tell you one thing. I have never run into a desktop system here in Taiwan that hasn't had either a CD or DVD drive installed. The only systems lacking optical drives have been laptops and all of those have been capable of booting from USB.
I feel they way they do desktop selection in the install is just fine. As for partitioning, who knows?
FWIW, 2 of my customers here in Taiwan are "market research" companies. I'm very good friends with the President of one company and a principle in the other. When I mentioned "web inquiry" or "web polls" for gauging customer needs they both just basically start rolling on the floor laughing.
That being said, you may have a good point when it comes to USB support. I have not taken the time to look into it since I've not needed it. But, I recently needed to install Win7 on a laptop whose DVD is dead. MS has a "Windows 7 USB/DVD download tool". I pointed it to the ISO of a Win7 DVD and it created a bootable USB. I didn't read, or need, any documentation. If there isn't such a tool for Fedora, then it would be nice to have one. If there is one, but not documented, then it needs to be...preferably by someone who has a need and experience in using it.
On Mar 23, 2014, at 2:36 AM, Ralf Corsepius rc040203@freenet.de wrote:
On 03/23/2014 08:56 AM, Frank Murphy wrote:
On Sat, 22 Mar 2014 18:29:02 +0000 Timothy Murphy gayleard@eircom.net wrote:
Ed Greshko wrote:
Does anybody nowadays actually burn CDs or DVDs?
Yes, because unlike USB-sticks, I am archiving them and resort to using them as "ready-to-use/off-the-bookshelf" media, e.g. in cases of "emergencies".
Offtopic but for what it's worth pretty much all dye based DVD is effectively disqualified from archiving because of the massive variance in errors, with the initial error rates (immediately after writing) busting the relevant standard (ISO/IEC29121). To even consider it requires ISO/IEC 10995 discs, a dedicated archival recorder, periodic testing strategy to know when to migrate data to new discs, and proper storage conditions (low temperature and humidity).
Stamped DVD (packaged movies and games) is a whole different animal, they're reliable short of microwaving them, but are impractical for backups.
On Mar 23, 2014, at 5:12 AM, Tim ignored_mailbox@yahoo.com.au wrote:
Just recently I had to boot from USB to install onto a laptop with no optical drive. I found four different methods of putting a Linux installer onto a USB. One of them involved Windows, so that was out. Two of them were Linux programs for writing installers to the flash drive. One required adding things to a current install that I didn't want to do. The other was less painful, and worked for writing one installer to a USB, but not for another (I tried Fedora and Ubuntu on this laptop). The last was plain dd if=/install.iso of=/def/flashdrive (I'm paraphrasing), and that worked for the one that didn't work the other way.
All of which was more hassle than burning a disc, and was actually slower than burning a disc for one of the methods (I don't recall which, but I tend to think it was the dd method).
The slowness of USB writing is probably the block size. I have a Kingston USB that writes at 20MB/s if the block size is 256KB or larger. At the default of 512 byte block sizes, it's abysmal, not even 1MB/s. Maybe 250KB/sec? A Lexar USB stick writes at 1MB/s with 512 byte block sizes, and maxes out at 4MB/s. Point being the media makes a big difference, but they all write slowest with 512 byte block sizes so I'd think there's possibly some room for optimization here.
Chris Murphy
On Mar 23, 2014, at 6:27 AM, Ian Malone ibmalone@gmail.com wrote:
On 22 March 2014 16:40, Liam Proven lproven@gmail.com wrote:
On 22 March 2014 03:54, Chris Murphy lists@colorremedies.com wrote:
Are you asserting that people who bitch and whine about the installer are entitled to a monopoly on mocking hectoring tone; and unhelpful, annoying, unproductive engagement?
I don't think any of us should do it, ideally. But perhaps in gentle chiding of someone who could have answered their question with Google in 10sec, then OK.
When someone is unable to even install the whole OS, no. Inappropriate.
It is actually a complex layout.
I beg to differ. It is not /trivial/ but it is not complete.
Most of the world's installers can't deal with what you just described.
Factually incorrect. Windows 7, Ubuntu 13.10, Debian 7, Crunchbang and Elementary OS all had no problems.
On my desktop PC, I have a similar layout with Windows 8, Mac OS X 10.6 and Ubuntu 13.10. Again, no problems at all.
The #1 OS install today is software restore [...]
In my extensive experience of OSes going back to when I entered the business in 1988, following about 6-7y as a hobbyist, this is incorrect. You're describing one OS, principally - modern Windows.
So, I read "#1 OS install today" as the installer for the #1 OS. There were a couple of other possible interpretations, but seeing #2 as the Windows retail/upgrade and #3 as Mac that made the most sense.
It was poorly worded on my part. It's a ranking of the most common install experience taking all of consumer computing in aggregate. The true #1 experience is actually pre-installed OS, but if you install an OS the most common experience is actually a re-installation via software restore. Windows, iOS, Android use this, and in ancient history Mac OS once did too. So the #1 ranking is multi-OS, it's the "restore" method. You have no choices other than to do it or not do it.
It sounds like either you've been spectacularly unlucky installing Fedora or have some particular requirement (hardware, install setup) that it is failing to deal with.
I suppose that's a possible explanation, no matter how inexplicably improbable it seems this could occur for ten years, presumably on different hardware over that decade, while the same installer used by CentOS worked. But I don't know how bad luck explains the lack of evidence and curiosity.
Chris Murphy
Timothy Murphy gayleard@eircom.net writes:
One of the weaknesses of Fedora, in my view, is the apparent lack of interest in what users actually want or need.
+1
For example, what proportion of users choose KDE or Gnome, or some other Desktop Environment? Does anybody know? Suppose it turned out that far more use KDE than Gnome. Would that not suggest that KDE should be made the default?
I`d rather see choices. There still can be a default, and it doesn`t matter what it is: Just give the user choices when installing and afterwards. One user might want to install gnome and kde and xfce and whatever else there is to try them out. Another one may already know that they want gnome. I already know that I don`t want a graphical login and fvwm. Where is that choice? Why do we have to somehow figure out how to get what we want only after the install, when it`s kinda too late?
Just think of all the time and resources wasted with downloading things we don`t need ...
Again, how many are installing on a machine that is already partitioned? Suppose (my guess) it is something like 90%. Would it not be more rational in that case to assume as default that users would prefer to use that partitioning?
No.
The capacity required for /usr has increased over time. Same goes for / or /boot. Some people may want to keep an existing system and need an option in the installer to shrink some partitions to make room. Some people may install on a (new) computer they were unable to get without something pre-installed, and they don`t like whatever partitioning was used for that. Others may have upgraded their hardware and have added a few disks, and now they want to use the additional capacity and need to move their /home partition, and this kind of upgrade and change may be the reason that they think they can as well re-install. Others plugged in a(n old) disk they had sitting around because they want to use it to try out Fedora, and the partitioning of that disk has been made a long time ago for, let`s say, Debian.
You can as well --- or even more reasonably --- assume that it is most unlikely that someone wants to use an existing partitioning as the default when installing.
Bill Oliver vendor@billoblog.com writes:
On Sun, 23 Mar 2014, Timothy Murphy wrote:
Tim wrote:
[snip] But all this is pure speculation. One of the weaknesses of Fedora, in my view, is the apparent lack of interest in what users actually want or need.
[snip]
I don't think it's a total lack of interest. I think that it's an issue of prioritization. From what I've read, the purpose of Fedora is to use the open source community to examine and debug things for potential inclusion in the Red Hat Enterprise products.
The mission statement says otherwise.
Accordingly, it is our "job" to get stuck with stuff we don't like and things that are not quite ready for prime time.
Wouldn`t it be more efficient to listen to what users want than it is to do lots of work to throw things in to see whether the users like them or not?
My impression is that Red Hat is *very* interested in what we put in bugzilla, and *is* interested in usability and preference issues from us -- but that the debugging part is first, usability second, and preference third in the priority list.
Bug reports are surely important, and the ones I made have been attended to. They are bug reports, though, i. e. you can say things like "I wanted to do this, so I tried to do that, and instead of getting X, the program crashed." That is a limit.
Some things you can only learn from discussions like this or from actually asking the users, and bug reports aren`t suited for that.
A reason for perceiving the apparent lack of interest may be that the makers of Fedora don`t appear to be present here. It is here where people ask questions, including usability issues, and come up with ideas and can say what they would like to see.
Using Fedora is like getting a big box of presents every six months sent to me by a rather absent-minded elderly aunt, who can't quite remember my tastes. The good part is that it is a big box full of neat stuff. The bad part is that some of it got broken in the mail and some of it I just don't care for. But I throw the stuff I don't like away and enjoy playing with the stuff I like.
That`s a good analogy :) Is it supposed to be like this?
Let`s see: Your aunt is running a bug tracking system to keep track of what got broken in the mail so she can resend what got broken, plus a mailing list so their grandsons and granddaughters can chat about what they got this time to figure out whether it is actually broken or not.
If she had known that she didn`t need to throw in this or that, wouldn`t that be easier for everyone?
Ian Malone ibmalone@gmail.com writes:
On 21 March 2014 15:19, lee lee@yun.yagibdah.de wrote:
One thing Fedora shines with (so far) is reliability, and reliability is one of the requirements I have. I have been using Debian for almost twenty years until they messed up badly with their brokenarch. Doing that put Debian out of the question once and for all because they failed that requirement miserably beyond believe.
Please do not make the same mistake with Fedora. Switching to another distribution is a painful process.
I'm kind of interested to know what your use case is that makes switching distros more painful than fixing a broken install. The only guess I can make is multiple systems.
That`s the wrong guess. One thing is the installation itself during which I need to keep my data save. That means I have to unplug all disks put the two I`m installing on. Then my internet connection is slow, and it`s a lot to download. Then I have to set up everything I have working now again. Backing up configuration files helps, yet every distribution has their own way of doing things, so I there is plenty to figure out. Then I need to get rid of stuff that gets installed which I don`t need or want to have. Over time, I have to install things I need again, which involves figuring out which packages might provide it. Depending on which distribution it is, I have to learn how to use a different package management, and I have to find out how something works when it`s different from how it was before, which I do not know to begin with and also need to figure out first. Configuration files are stored at different places with different distributions, some things are configured very differently, and some packages I need may not be available.
When switching distributions, it takes at least half a year to get back to where I was before.
To give you a silly example: When I was using Debian, I had eximstats running, coordinated with logrotate. In over a year I haven`t figured out yet how to do that with Fedora. It`s not really important, and it seemed not even to be available, and I had more important things to do. Only recently I happened to find that it`s suddenly installed, so how do you set it up?
Another example: I need squid 2.7 because I need a feature that still hasn`t been ported to the current version. Squid 2.7 is not in Fedora, and it was difficult to get it to work because I had to learn how to write some startup file for systemd and to find a way to somehow trick unknown things Fedora has which prevented squid from being able to have a log file and even kept deleting directories I made. Squid 2.7 is in Debian, so you simply install it, adjust the configuration file and you`re done.
And more examples: When I log in as another user, that other user has no sound. I`ve been asking on this list quite a while ago and didn`t get an answer. Any idea how to fix that? It never was a problem with Debian.
Debian didn`t use pulseaudio, so the sound worked just fine. Until F19, pulseaudio used to randomly crash, which disabled all sound. It takes time to figure out things like that and that you have to restart pulseaudio to get the sound back.
The package for fvwm in Debian was ancient, so I had a self-compiled version. That is no longer needed with Fedora: just another detail that needed to be looked into. Xterm, which worked fine with Debian, would blank out, so I couldn`t use it for quite a while: just another detail.
Where is xcolorsel? I sometimes need that, and I still haven`t found a replacement or a way to get it with Fedora.
Where is cinelerra? I wanted to learn how to use it, but I can`t get it with Fedora :(
It`s hundreds of little details that need to be looked into, and many things you can`t just do as you used to. They are either done differently, or it`s something you set up so long ago that you forgot how you did it --- or something just doesn`t work anymore or is not available.
So switching distributions is a major PITA. It`s sure fun to try out different things and learn new stuff --- if that is what you want to do. I`m doing it, and it`s fun, but I need my stuff working in the first place and not to get it in the way of what I want to do.
Even after over a year, I don`t have everything back I had with Debian. Fedora has only maybe half the packages Debian has.
Perhaps Feodra.next can catch up in that regard?
I have always wondered how people manage to create packages, for Debian
[...]
find out how to do that and that`s where it ends: It`s just too difficult.
Instead, you put your software on github.
If you see no value in packaging I'm surprised you aren't using a roll-your-own distro instead. It sounds like that would be a much better match for your requirements of control over everything on the machine and a minimalist system.
Not at all. Like I said in another post (or at least I think I did): I love packages. Without packages, you have chaos and the problem of having to piece together what you need and of having to somehow monitor everything you use for changes.
And I don`t want a "minimalist system" either. That is very different from not wanting things I don`t need or don`t want.
Think of cinelerra, for example. I suppose if you don`t want a "minimalistic system", you won`t want to have it. I want to have it and want to simply install a package that provides it.
Think of fvwm: You may think it`s a "minimalistic" window manager. It is not, it`s actually the most powerful and most versatile WM I have ever seen, and it`s easy to configure to do exactly what you want. Do you know of any other WM that even comes close? I`ve used it about twenty years ago and tried many others in between and I came back to it simply because there is nothing better. It`s the opposite, if there`s such a thing, of "minimalistic".
Think of emacs: You may also think it`s "minimalistic", and when you look closer, you may find it`s the opposite. I have used it about twenty years ago and many other editors in between, and I came back to it: There is nothing better. (I don`t get along with vi.)
Now think gnome: You may think it`s not "minimalistic". Well, it is, it doesn`t let me do anything and totally gets into my way and is not configurable, yet it runs all kinds of stuff I don`t need or want. I tried it out every now and then over the years and never found any use for it.
Think KDE: I actually used it for a short while, not even long ago, but it had so many bugs and quirks and so many things not working and wasn`t really configurable, either, so I ditched it when it was updated and things got even worse. To do all kinds of things I don`t need or want, it slowed my computer down noticeably. It`s "minimalistic".
Think LaTeX: You may think it`s "minimalistic" compared to Libreoffice. Pfff. I used it about twenty years ago because there wasn`t anything better. There still isn`t --- and the LaTeX sources from back then can still be worked with and printed with no problem. I still have them. I tried Libreoffice not long ago and it was unable to do a simple thing they call mail-merge, it would just crash. It might have been possible to learn how to program it and to achieve what I wanted otherwise, but I stumbled over one bug after the other. Libreoffice is "minimalistic". I don`t even have it installed. Even if it worked, it would not be worthwhile to learn how to program it because it`ll be gone in twenty years or have changed so much that you`d have to start learning all over again.
Don`t get me wrong. LaTeX isn`t a good solution for everything, there are WMs that have features that fvwm doesn`t and doing spreadsheets in emacas, though you can, is awkward. I also have nemo running because it is sometimes useful.
Hm. I was about to say that nemo uses ridiculous amounts of memory and looked. Guess what: It`s not running anymore, it has crashed. Minimalistic ...
Anyway, I`m not into having a "minimalistic" system. I like efficiency and stuff that works, is well thought out and does what I want without unnecessary fuss. That kind of software usually happens to be extremely powerful. I don`t like having stuff which I don`t need and wasting resources on it.
I don`t know what would be better than packages. I wish there were more packages and am only saying it`s too difficult to make them. Perhaps there`s good reason for it, perhaps it can be made easier. Someone who knows how to make them might be able to tell.
On Sun, 23 Mar 2014 16:35:20 +0100 lee wrote:
One of the weaknesses of Fedora, in my view, is the apparent lack of interest in what users actually want or need.
+1
The distro that actually seems to care (at the moment) about users is Linux Mint (though I have no doubt that, like all open source projects, it will eventually be taken over by a cadre of "helpful" zealots who will yank the steering wheel in different direction because stupid users don't realize what they actually need :-).
I use Fedora primarily because it is the best distro to use as an early warning system to find out what nonsense will land in RHEL in the future, and at work we need our software to run on RHEL, so Fedora is sort of like the Canary in the coal mine.
On Sun, 23 Mar 2014 16:15:03 +0100 lee wrote:
Think of fvwm: You may think it`s a "minimalistic" window manager. It is not, it`s actually the most powerful and most versatile WM I have ever seen, and it`s easy to configure to do exactly what you want.
No only that, but it works the same way all the time as long as you keep your same .fvwmrc files. Imagine that: You don't have to re-learn the user interface on every release.
On 22 March 2014 16:55, Tom Horsley horsley1953@gmail.com wrote:
On Sat, 22 Mar 2014 16:40:36 +0000 Liam Proven wrote:
Meantime, for further Fedora eval, it's going in a VirtualBox. Sad, but that's all it seems able to handle.
Actually, that's the key to installing on disk in a sensibly created partition layout: Install in a virtual machine 1st, then copy the VM disk image to a real disk and fiddle with the grub config and fstab. I install that way all the time now since the new installer appeared, and find it actually make things lots more convenient since I don't have to have the machine down while I'm doing the install.
[Nod] I might try that. Assuming that I actually like the OS enough to want to commit it to the bare metal, of course. ;-)
But if this is SOP, then that to me says that the installer is broken to a laughable degree.
For instance, many OSes can't handle the PC primary+secondary partitioning scheme. This means that they need their own primary partition, which is a *major* limiting factor if you have a multiboot PC - you only get 4 primaries per drive, and if you need logical partitions, then one of those primaries must be an extended partition, so you only get 3. Primaries are a precious, scarce resource, and that means that OSes that demand primaries - *BSD, Solaris, BeOS/Haiku, etc. - are not good citizens when it comes to multiboot machines.
But Linux has always been the opposite. It's the eternal secondary OS, running on machines designed for other OSes - primarily Windows, of course, but also on hardware designed to run Mac OS X, Solaris, AIX, HP-UX, etc. etc. If there's any OS that runs on more platforms than Linux it can only be NetBSD and it is a real minority OS for a handful of users.
Linux copes with /everyone's/ partitioning systems. Apple, Sun, PC, GUID, it doesn't care, it just works with it.
Fedora, it seems, does not. It wants things done its way, or perhaps as a secondary OS on a Windows machine. Otherwise, you're quite possibly out of luck.
For one of the big, famous, heavily-promoted "outreach" distros, one that is famous, I submit that this is not a good approach.
Just my 2¢'s worth!
On 22 March 2014 17:06, Patrick O'Callaghan pocallaghan@gmail.com wrote:
On Sat, 2014-03-22 at 16:40 +0000, Liam Proven wrote:
As I have said previously, I have /never/ successfully installed Fedora on actual hardware since v1.0 shipped in, what was it, 2003? I have installed Haiku, Aros, FreeBSD, PC BSD, dozens of Linux distros, Windows 2 through 8, SCO Xenix, SCO Unix, OpenSolaris, OpenVMS, FreeDOS, DR-DOS, MS-DOS, PC-DOS, OS/2 1 through eComStation 2, MacOS 6 through OS X 10.9. I am *not* a newbie and I am *not* an inexperienced inexpert fumbler.
I know you've said this before but I still find it astonishing. I've used Fedora since FC1 and have never failed to install it or update it on real hardware (and not always the same real hardware). Once or twice I've had issues with the layout (in fact I had some with F20 when it decided I wanted a BTRFS subvolume spread over two disks) but they've been the exception. I do agree that the installer needs to be clearer than it is, though I find F20 better than F19.
I'm not questioning your experience, I just find it remarkably different from my own.
I've had similar experiences myself - where stuff that works really easily and simply for me just doesn't for others, or collapses or explodes in legion bizarre errors and so on.
The thing is, edge cases like this are really informative. It's really hard to trace an obscure intermittent bug and it's really hard to file a useful bug report when the bug is "it doesn't work for me" and you can't really add more data than that.
But if you have a whole bunch of users saying "this doesn't work for me" or "I find this really hard" or "this program can't give me the options I want", then that to me seems like a big obvious signal that something is badly wrong and needs examination and possibly re-consideration.
It's not much use for tracing a particular bug - chances are that for everyone who's failed to install an instance of F20, the reason is different. There's no one bug to file.
In this case, for instance, I'd point the accusing finger at the installer developers' assumptions that:
[a] on a system with 2 disks, either of those disks can be usefully analysed by the install program in isolation. That's an invalid assumption and the result is a program that can't handle many common instances.
[b] that one will want to install the bootloader on the MBR of the drive with the root filesystem. That's an invalid assumption, too.
[c] that one will want to install a bootloader into an MBR and nowhere else. That's an invalid assumption.
[d] that on a system with multiple Linux installs, you can group sets of partitions by install - in other words, these ones belong to install #1, these to install #2, these to install #3. That's invalid for me - I share /home and swap between all my installs. Only the root FS is unique to each. That breaks F20's installer.
These are not bugs. These are design errors, where someone has assumed that their personal preconceptions are universal truths.
On 22 March 2014 17:23, lee lee@yun.yagibdah.de wrote:
Let`s say you install to a software RAID-1 --- which is minimum requirement for anything to put data on --- made from two disks, with encrypted partitions (as usual /, /usr, /home, /tmp, /var, /usr/local, and a swap partition). You want to have these partitions in a particular order on the disks, i. e. swap at the beginning because chances are it`s faster, then /usr, /var, /tmp, /usr/local and /home, in that order.
That`s nothing complicated, either, and I don`t think that`s possible with Fedoras installer. Or is it? And if it is, how long does it take to do the partitioning?
[Blinks]
Wow. Now, y'see, that's something I'd consider wildly exotic and weird. I haven't put /usr on a separate partition since the 1980s when I was trying to build 20-user systems with 20-40MB hard disks.
I never separate out /tmp or /var or /usr/local - I only ever use / and /home basically. I might split off /var on a server but I'd need a remarkably persuasive use case, and on servers, I use extra-stable distros without GUIs, not something like Fedora.
But this just illustrates the breadth of scenarios a successful installer must cope with!
On 03/23/2014 04:12 AM, Tim wrote:
Just recently I had to boot from USB to install onto a laptop with no optical drive. I found four different methods of putting a Linux installer onto a USB. One of them involved Windows, so that was out. Two of them were Linux programs for writing installers to the flash drive. One required adding things to a current install that I didn't want to do. The other was less painful, and worked for writing one installer to a USB, but not for another (I tried Fedora and Ubuntu on this laptop). The last was plain dd if=/install.iso of=/def/flashdrive (I'm paraphrasing), and that worked for the one that didn't work the other way.
If you haven't already, you may want to take a look at unetbootin. Unlike Fedora's Live USB Creator it doesn't care what distribution you're putting on the drive.
On 22 March 2014 18:25, Ed Greshko ed.greshko@greshko.com wrote:
Make sure you file a bugzilla.
How/why? It's not a bug.
Talking about it on this list will not result in any action.
Won't it? Why not? Are there multiple non-overlapping user communities for Fedora, then? A problem is a problem; where it is reported shouldn't matter, should it?
Of course, you'll have to be able to consistently repeat the problem and be willing to document it well.
No can do. The bug is "won't install on my computer". That's not useful info and that's why I would not normally consider filing it.
FWIW, for problems such as this one I just happen to have spare disks that aren't bad. They were just replaced for ones of larger capacity. I understand you may not be in the same position.
I do not understand this paragraph at all.
On 22 March 2014 18:29, Timothy Murphy gayleard@eircom.net wrote:
How old are they?
May I just point out at this instance that the current versions of VMware and VirtualBox both cannot boot from USB?
It is an error to assume that all modern hardware can. It can't.
Tom Horsley horsley1953@gmail.com writes:
On Sun, 23 Mar 2014 16:35:20 +0100 lee wrote:
One of the weaknesses of Fedora, in my view, is the apparent lack of interest in what users actually want or need.
+1
The distro that actually seems to care (at the moment) about users is Linux Mint
They use Debian packages, don`t they? That is something I wanted to get away from.
I use Fedora primarily because it is the best distro to use as an early warning system to find out what nonsense will land in RHEL in the future
And otherwise you`d be using Mint?
I`m somewhat surprised that the feeling of apparent desinterest of the makers of Fedora in what its users think seems kinda widespread under its users. Perhaps it`s a wrong impression; if not, it may be something for Fedora.next to address.
Many users probably don`t care as long as Fedora works for them --- and when it doesn`t anymore, they just pick something else?
That surely wouldn`t bring Fedora anywhere near a position in which it should contemplate leading the advancement of FOSS ... unless what users of FOSS think is indeed irrelevant.
Tom Horsley horsley1953@gmail.com writes:
On Sun, 23 Mar 2014 16:15:03 +0100 lee wrote:
Think of fvwm: You may think it`s a "minimalistic" window manager. It is not, it`s actually the most powerful and most versatile WM I have ever seen, and it`s easy to configure to do exactly what you want.
No only that, but it works the same way all the time as long as you keep your same .fvwmrc files. Imagine that: You don't have to re-learn the user interface on every release.
Not only that, you only need to learn once how to set it up the way you want it.
Now someone tell me how I get virtual desktops with gnome or kde to which I can switch by just moving the mouse pointer over the edge of the screen? How do I get a pager as I have with fvwm with gnome or kde? Where is the configuration file for defining my key bindings, menues, window decorations ...?
I once tried a window manager that would show some sort of cubes when you wanted to switch desktops. It somehow required a huge amount of resources to do that. People were really excited about it. I found that it looks nice and totally gets into the way because I`d be looking at those stupid cubes when I wanted to switch desktops and they won`t really let me and were only one dimensional (i. e. you could turn them left and right but not up and down or move several around on your screen to pick a desktop from one of them) and served no purpose. I forgot how it was called ... compitz maybe?
With fvwm, I just move the mouse pointer over or press AltGr and an arrow key, and I`m on the next desktop without any delays or something getting into my way. It even does a better job with tiling than i3 does: It just does it, without having you try to figure out how to get your windows arranged and without the confusing containers which never really do what you want, and I can have sticky floating windows which you can`t have with i3 ...
Can gnome or kde do tiling? With sticky floating windows? I have that on one of the virtual desktops for two seamonkey windows; almost everything else is full screen.
Try full screen with gnome. You can`t, you have to move and resize your windows all the time and move them off half off-screen and they are still in the way. I hate that. How can all these fancy "desktop environments" be so incredibly far behind (a twenty year old window manager)?
How did that go? Fedora wants to lead the advancement of FOSS. Then why isn`t fvwm the default window manager? It`s the most advanced one you can get. It`s perfect for Fedora.next.
Then they need to turn around at once! Leaving users in the dark about what`s going on makes things more difficult for them. Taking away choices limits the use of the software to the point where it eventually becomes unusable.
Good design, clarity, good documentation and letting the user know what choices there are are required when you want to achieve ease of use.
I don't think the installer is as bad as most people on this list are complaining about; it's certainly not leaving any users in the dark. A large majority of the information from the older installer is still there, it's just up to the user to seek it out.
In essence the installer went from a hardcore presentational format to a more laid back format - a shift every OS has consciously made in the last decade.
On 23 March 2014 15:15, lee lee@yun.yagibdah.de wrote:
Ian Malone ibmalone@gmail.com writes:
On 21 March 2014 15:19, lee lee@yun.yagibdah.de wrote:
And I don`t want a "minimalist system" either. That is very different from not wanting things I don`t need or don`t want.
Think of cinelerra, for example. I suppose if you don`t want a "minimalistic system", you won`t want to have it. I want to have it and want to simply install a package that provides it.
Think of fvwm: You may think it`s a "minimalistic" window manager. It is not, it`s actually the most powerful and most versatile WM I have ever seen, and it`s easy to configure to do exactly what you want. Do you know of any other WM that even comes close? I`ve used it about twenty years ago and tried many others in between and I came back to it simply because there is nothing better. It`s the opposite, if there`s such a thing, of "minimalistic".
Think of emacs: You may also think it`s "minimalistic", and when you look closer, you may find it`s the opposite. I have used it about twenty years ago and many other editors in between, and I came back to it: There is nothing better. (I don`t get along with vi.)
There is no-one on the planet who thinks emacs is minimalistic (and yes I use it fairly extensively, but only because I work a lot with systems where a good IDE would not help). With this and all your other examples you are talking about individual pieces of software. But what I mean is you've said you want specific things on the system that are not what other people would want and things they want are not what you want. On the Venn diagram of what's included in an install that makes life difficult.
Think LaTeX: You may think it`s "minimalistic" compared to Libreoffice.
I don't, we both seem to be making a lot of assumptions about what each other thinks, but I've tried to base mine on what you've said.
wasn`t anything better. There still isn`t --- and the LaTeX sources from back then can still be worked with and printed with no problem. I still have them.
Provided they only use packages that are compatible with the TeX distro you've got available.
I don`t know what would be better than packages. I wish there were more packages and am only saying it`s too difficult to make them. Perhaps there`s good reason for it, perhaps it can be made easier. Someone who knows how to make them might be able to tell.
Ultimately writing a basic spec file is pretty simple, if you can do the configure-make-install cycle it should be straightforward enough.
On 03/23/2014 12:08 PM, lee wrote:
compitz maybe?
Close: compiz. I use it with Xfce because I enjoy watching the eye candy. And, I use it (also with Xfce) on my laptop so that I can show all of the "gosh wow" effects to people who only know Windows. Then, I tell them that the laptop came with Vista and has never had a hardware upgrade and ask, "If Linux can do this, why can't Windows?" I've never yet gotten an answer.
On Mar 23, 2014, at 12:19 PM, Liam Proven lproven@gmail.com wrote:
The thing is, edge cases like this are really informative. It's really hard to trace an obscure intermittent bug and it's really hard to file a useful bug report when the bug is "it doesn't work for me" and you can't really add more data than that.
OK but what you've done is said what I've never heard anyone say, which is having never successfully installed Fedora to baremetal in 10 years. And on top of it, you haven't stated the nature of the failure, either the error messages you've gotten, or the logs the installer always produces from the moment it's launched.
But if you have a whole bunch of users saying "this doesn't work for me" or "I find this really hard" or "this program can't give me the options I want", then that to me seems like a big obvious signal that something is badly wrong and needs examination and possibly re-consideration.
Define "whole bunch", how this is ascertained, and how the assertion applies to Fedora.
It's not much use for tracing a particular bug - chances are that for everyone who's failed to install an instance of F20, the reason is different. There's no one bug to file.
I don't even know what you mean by "failed to install" because there are several ways that can manifest but in any case the installer always writes log files from the moment it's launched, figuring out the problem isn't difficult especially when it's reproducible.
In this case, for instance, I'd point the accusing finger at the installer developers' assumptions that:
[a] on a system with 2 disks, either of those disks can be usefully analysed by the install program in isolation. That's an invalid assumption and the result is a program that can't handle many common instances.
I can't tell whether you're referring to Fedora's installer or not. But it doesn't work this way.
[b] that one will want to install the bootloader on the MBR of the drive with the root filesystem. That's an invalid assumption, too.
Why, and what alternative is valid?
[c] that one will want to install a bootloader into an MBR and nowhere else. That's an invalid assumption.
A default bootloader had to be chosen, and GRUB2 is what was chosen. The boot.img portion can only go in the MBR, and the core.img is too big to fit in a VBR so it's usually put in the MBR gap (or BIOS Boot on GPT disks). There really isn't a work around for this given the design of GRUB2. If you want to install a bootloader to a VBR then a different bootloader selection is needed, such as extlinux. Fedora supports this as a hidden option, right now it's too much effort to give it equally visible support. I'm not aware of any other distribution that even supports two bootloaders. All kernel updates support either GRUB or extlinux (and others).
[d] that on a system with multiple Linux installs, you can group sets of partitions by install - in other words, these ones belong to install #1, these to install #2, these to install #3. That's invalid for me - I share /home and swap between all my installs. Only the root FS is unique to each. That breaks F20's installer.
Define "breaks" please produce log files for these install attempts. I have done what you suggest and it doesn't break for me, it installs and uses a pre-existing /home and swap just fine.
These are not bugs. These are design errors, where someone has assumed that their personal preconceptions are universal truths.
Flawed logic. The design decisions aren't going to mesh with everyone's requirements. Multibooting on BIOS computers is fundamentally flawed besides not being standardized at all, and therefore there is no one correct way to arrive at a multiboot system, and no one correct way to interpret a multiboot system and always support installing to multiboot systems.
Chris Murphy
On 03/23/2014 12:34 PM, lee wrote:
I`m somewhat surprised that the feeling of apparent desinterest of the makers of Fedora in what its users think seems kinda widespread under its users. Perhaps it`s a wrong impression; if not, it may be something for Fedora.next to address.
Well, Fedora is Gnome-centric, and many of us get the same impression about the Gnome devs.
On 23 March 2014 19:08, lee lee@yun.yagibdah.de wrote:
Tom Horsley horsley1953@gmail.com writes:
On Sun, 23 Mar 2014 16:15:03 +0100 lee wrote:
Think of fvwm: You may think it`s a "minimalistic" window manager. It is not, it`s actually the most powerful and most versatile WM I have ever seen, and it`s easy to configure to do exactly what you want.
No only that, but it works the same way all the time as long as you keep your same .fvwmrc files. Imagine that: You don't have to re-learn the user interface on every release.
Not only that, you only need to learn once how to set it up the way you want it.
Now someone tell me how I get virtual desktops with gnome or kde to which I can switch by just moving the mouse pointer over the edge of the screen? How do I get a pager as I have with fvwm with gnome or kde? Where is the configuration file for defining my key bindings, menues, window decorations ...?
kde: System Settings | Workspace Behaviour | Screen Edges | Other Settings | Switch desktop on edge (Disabled/Only when moving windows/Always Enabled). Also, System Settings | Workspace Appearance for appearance changes, Shortcuts and Gestures for...
I once tried a window manager that would show some sort of cubes when you wanted to switch desktops. It somehow required a huge amount of resources to do that. People were really excited about it. I found that it looks nice and totally gets into the way because I`d be looking at those stupid cubes when I wanted to switch desktops and they won`t really let me and were only one dimensional (i. e. you could turn them left and right but not up and down or move several around on your screen to pick a desktop from one of them) and served no purpose. I forgot how it was called ... compitz maybe?
Compiz. Which did compositing (and didn't really need massive resources so much as graphics hardware support). Had some nice and genuinely useful features which have not carried over to newer WMs which have adopted compositing.
Though the cubes themselves were pretty cosmetic it was the same technology also that also let you view all desktops live when switching.
With fvwm, I just move the mouse pointer over or press AltGr and an arrow key, and I`m on the next desktop without any delays or something getting into my way. It even does a better job with tiling than i3 does: It just does it, without having you try to figure out how to get your windows arranged and without the confusing containers which never really do what you want, and I can have sticky floating windows which you can`t have with i3 ...
Can gnome or kde do tiling? With sticky floating windows? I have that on one of the virtual desktops for two seamonkey windows; almost everything else is full screen.
KDE does not do true tiling (Compiz did). It does do pining, which is what the pin button at the top left of each window does. And keyboard shortcuts for desktop changes (thought the default is Ctrl+F1-4 rather than ctrl+alt and arrow).
On Mar 23, 2014, at 12:24 PM, Liam Proven lproven@gmail.com wrote:
On 22 March 2014 18:25, Ed Greshko ed.greshko@greshko.com wrote:
Make sure you file a bugzilla.
How/why? It's not a bug.
How do you know? You haven't filed a bug report, you haven't presented any logs to anyone to find out if what you're experiencing is in fact intended behavior. You haven't even stated the nature of the installation failure.
Talking about it on this list will not result in any action.
Won't it? Why not? Are there multiple non-overlapping user communities for Fedora, then? A problem is a problem; where it is reported shouldn't matter, should it?
It will not. Because this is the user to user forum. To cause a design change in the installer means posting to the Red Hat anaconda-devel@ list. To cause a design change specific to Fedora means posting to the Fedora devel@ list. Bugs, requests for enhancement, and even perceived clearly articulated design flaws should have bug reports filed. Pretty much always when an installer bug is filed, the installer logs should be attached (at least program.log and storage.log).
Yes it matters.
Of course, you'll have to be able to consistently repeat the problem and be willing to document it well.
No can do. The bug is "won't install on my computer". That's not useful info and that's why I would not normally consider filing it.
You're seriously acting like this is your first day at the rodeo. You are supposed to state what did happen, including any error messages received, and how this differs from your expectation of what should have happened. The installer always keeps logs in /tmp - so go try a Fedora baremetal install and when it "won't install" and you're stuck, post the /tmp/anaconda.log, /tmp/program.log, and /tmp/storage.log.
You can even fpaste them directly from the shell, just fpaste <filename> and then note the URL and post it.
Chris Murphy
On Sun, 23 Mar 2014, lee wrote:
[snip] If she had known that she didn`t need to throw in this or that, wouldn`t that be easier for everyone?
Not necessarily. Sometimes neither that absent-minded aunt nor the avaricious nephew know what is most pleasing until it's taken out and played with. Much of what we consider very important today was considered stupid when it first came out.
I've had good friends who worked at Xerox PARC back it its heyday, others who worked at Bellcore, and another who worked at that ivory tower in Redmond whose name I forget. They all tell the same story of coming out with neat stuff that the mother ship thought would never catch on. Which is why those neat places almost always end up getting closed down or changed -- not because they don't come up with great ideas, but because nobody in corporate knows what to do with them, or because they are disruptive and corporate doesn't want to change how they do things (e.g. Kodak).
And it's not just the corporate heads that get it wrong. Sometimes consumers don't take to an idea before they taste the real thing.
So, for new stuff, maybe you just don't know whether you "need" to throw something in until you do throw it in, keep doing it for awhile, and see if it catches on.
billo
On Sun, 2014-03-23 at 18:24 +0000, Liam Proven wrote:
On 22 March 2014 18:25, Ed Greshko ed.greshko@greshko.com wrote:
Make sure you file a bugzilla.
How/why? It's not a bug.
If it doesn't work as it's supposed to, it's a bug.
Talking about it on this list will not result in any action.
Won't it? Why not? Are there multiple non-overlapping user communities for Fedora, then? A problem is a problem; where it is reported shouldn't matter, should it?
Seriously? Yes, there are multiple partially-overlapping communities. Many of the developers don't read this list. Many of the people on this list don't read the Test list. If you want to document a bug then use Bugzilla.
poc
Bill Oliver vendor@billoblog.com writes:
On Sun, 23 Mar 2014, lee wrote:
[snip] If she had known that she didn`t need to throw in this or that, wouldn`t that be easier for everyone?
Not necessarily. Sometimes neither that absent-minded aunt nor the avaricious nephew know what is most pleasing until it's taken out and played with.
That can be an important thing to do. It requires listening to the users and informing and preparing them beforehand, and you have to have something available to fall back to, or a fix, in case the new toy doesn`t work out. If you leave them suddenly stranded like Debian did, they will leave you behind.
Much of what we consider very important today was considered stupid when it first came out.
Like?
Joe Zeff joe@zeff.us writes:
On 03/23/2014 12:34 PM, lee wrote:
I`m somewhat surprised that the feeling of apparent desinterest of the makers of Fedora in what its users think seems kinda widespread under its users. Perhaps it`s a wrong impression; if not, it may be something for Fedora.next to address.
Well, Fedora is Gnome-centric, and many of us get the same impression about the Gnome devs.
Well, I don`t use gnome, though I understand what you mean. Still when makers of software X appear desinterested, that doesn`t mean that makers of software Y need to or should appear the same way.
Besides, Fedora.next shouldn`t be limited to gnome ...
Joe Zeff joe@zeff.us writes:
On 03/23/2014 12:08 PM, lee wrote:
compitz maybe?
Close: compiz. I use it with Xfce because I enjoy watching the eye candy.
There`s nothing wrong with it when you like the eye candy. It`s good to have choices :)
And, I use it (also with Xfce) on my laptop so that I can show all of the "gosh wow" effects to people who only know Windows. Then, I tell them that the laptop came with Vista and has never had a hardware upgrade and ask, "If Linux can do this, why can't Windows?" I've never yet gotten an answer.
Well, yes, I don`t understand either why there isn`t a useable WM for it. I only touch it when I get payed for it, so it doesn`t really matter when three quarters of the time is wasted with moving windows around and half the screen space is covered by their decorations.
Macs are even worse, though I never looked if there might be a replacement for their WM. Perhaps you can have fvwm on a Mac?
"Powell, Michael" Michael_Powell@mentor.com writes:
Then they need to turn around at once! Leaving users in the dark about what`s going on makes things more difficult for them. Taking away choices limits the use of the software to the point where it eventually becomes unusable.
Good design, clarity, good documentation and letting the user know what choices there are are required when you want to achieve ease of use.
I don't think the installer is as bad as most people on this list are complaining about; it's certainly not leaving any users in the dark.
It doesn`t give you choices. It leaves you in the dark about that it is somehow possible to use an non-gui installer and to do a minimal install. It leaves you in the dark about what exactly happens when you do the partitioning and with trying to figure out how get the partitioning you want.
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
A large majority of the information from the older installer is still there, it's just up to the user to seek it out.
I don`t know anything about "the older installer" or how to seek out information about it; I didn`t even know that there is an "older installer". The first Fedora installer I used was the one with F17.
In essence the installer went from a hardcore presentational format to a more laid back format - a shift every OS has consciously made in the last decade.
I don`t know what you mean. The Debian installer got more options and some more clarity which was an improvement. Otherwise it didn`t change, you just do country and keyboard setup --- which is missing in Fedoras installer, there was no way to tell it that I have a German keyboard --- network setup if you don`t use DHCP, partitioning, a bit of package selection if you want to, and then it installs.
It`s easy and straightforward as it used to be for the last twenty years, and I never had trouble using it. Why suddenly make installing such a PITA like Fedoras installer does?
Liam Proven lproven@gmail.com writes:
On 22 March 2014 17:23, lee lee@yun.yagibdah.de wrote:
Let`s say you install to a software RAID-1 --- which is minimum requirement for anything to put data on --- made from two disks, with encrypted partitions (as usual /, /usr, /home, /tmp, /var, /usr/local, and a swap partition). You want to have these partitions in a particular order on the disks, i. e. swap at the beginning because chances are it`s faster, then /usr, /var, /tmp, /usr/local and /home, in that order.
That`s nothing complicated, either, and I don`t think that`s possible with Fedoras installer. Or is it? And if it is, how long does it take to do the partitioning?
[Blinks]
Wow. Now, y'see, that's something I'd consider wildly exotic and weird.
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
RAID isn`t exotic, either. Disks do fail, the only question is when, and I neither want to lose data, nor the hassle.
Installing on a laptop requires encrypted partitions. They can be stolen too easily.
I never separate out /tmp or /var or /usr/local - I only ever use / and /home basically.
I always use separate partitions. It has lots of advantages.
I might split off /var on a server but I'd need a remarkably persuasive use case, and on servers, I use extra-stable distros without GUIs, not something like Fedora.
/var can get full, and it`s written to, same goes for /tmp. How do you mount /usr read-only? Especially on a server, it`s a good idea to mount everything read-only that you can. When you have several disks, you can do your partitioning in such a way that you get better performance. Especially when you have a server, you may need a (pretty much) granted capacity on /var or /tmp to make sure it will continue to operate --- without separate partitions, your users may fill up the disks ...
Nowadays you may have SSDs which supposedly last longer when not written much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
Why wouldn`t you use different partitions? I can see it (and have done it) for when the available disk capacity is extremely limited, but otherwise it doesn`t make any sense and has nothing but disadvantages.
But this just illustrates the breadth of scenarios a successful installer must cope with!
It`s merely a reasonable standard thing to use separate partitions and a requirement to use RAID, and encrypted partitions for laptops, not something in any way unusual. Of course I expect an installer to handle that as well as using a single, unencrypted partition on a single disk.
And it`s not too difficult. The installer doesn`t need to do the partitioning, the user does it. The installer only needs to give the user a good tool to do the partitioning the user wants and let them use it. Good tools to do partitioning are already available, and the installer doesn`t need to re-invent the wheel in that.
Perhaps it even shouldn`t. Why force the user to learn how to use yet another partitioning tool they even rarely use unless they install Fedora all the time? Why not give them a choice, like either cfdisk or parted, then tell the installer what to do with each partition and let them switch between these until they are done --- or let the installer do whatever partitioning it wants, which means that all existing data on the disks will be deleted.
Ian Malone ibmalone@gmail.com writes:
On 23 March 2014 19:08, lee lee@yun.yagibdah.de wrote:
Tom Horsley horsley1953@gmail.com writes:
On Sun, 23 Mar 2014 16:15:03 +0100 lee wrote:
Now someone tell me how I get virtual desktops with gnome or kde to which I can switch by just moving the mouse pointer over the edge of the screen? How do I get a pager as I have with fvwm with gnome or kde? Where is the configuration file for defining my key bindings, menues, window decorations ...?
kde: System Settings | Workspace Behaviour | Screen Edges | Other Settings | Switch desktop on edge (Disabled/Only when moving windows/Always Enabled). Also, System Settings | Workspace Appearance for appearance changes, Shortcuts and Gestures for...
That`s way more complicated than EdgeScroll 100 100. And what about gnome? I looked for it and didn`t even find a way to adjust the number of virtual desktops (or only that), let alone focus follows mouse and moving over the screen edges.
Compiz. Which did compositing (and didn't really need massive resources so much as graphics hardware support).
People seemed to worry a lot about it.
Had some nice and genuinely useful features which have not carried over to newer WMs which have adopted compositing.
Features like? It doesn`t have them anymore?
Though the cubes themselves were pretty cosmetic it was the same technology also that also let you view all desktops live when switching.
Hm, why would I want to do that? I have currently 6x6 and they would be too small to see anything.
KDE does not do true tiling (Compiz did).
Fvwm doesn`t, either, but I have some entries in the menu that tile some windows on the current desk. It works much better than a tiling WM like i3.
It does do pining, which is what the pin button at the top left of each window does.
You mean sticky floating windows? I3 is really nice, but it doesn`t do that, and I sometimes need them ...
And keyboard shortcuts for desktop changes (thought the default is Ctrl+F1-4 rather than ctrl+alt and arrow).
Ah yes, and those didn`t always work because everything had a key binding and they would conflict with each other ... That doesn`t happen with fvwm, they just work.
Many default key bindings are very useless to me because I use my trackball with my left hand, and I have a German keyboard which has an AltGr key on the right rather than an Alt key there. So I need key bindings I can use with my right hand.
And neither with kde, nor gnome you can even have the scroll bars on the left side where they belong :( Kde is at least capable, but when you do that, the menu entries inevitably move over to the right where they don`t belong and it gets even more awkward than it already is. Most X11 apps do it just fine, seamonkey does it --- and I think emacs too, but I turned them off.
Note for Fedora.next: Please make the scroll bars finally configurable and provide a way to switch the key bindings so that they can be used with the right hand ...
Ian Malone ibmalone@gmail.com writes:
There is no-one on the planet who thinks emacs is minimalistic
I`m not so sure about that. There are even ppl who have never heard about emacs, and if they`d see it, they might very well think it is.
With this and all your other examples you are talking about individual pieces of software. But what I mean is you've said you want specific things on the system that are not what other people would want and things they want are not what you want. On the Venn diagram of what's included in an install that makes life difficult.
Each user has preferences. It is not too difficult to give them choices. What I want to have installed is different from what someone else wants to have installed --- but we don`t get to pick before the installation is already done.
wasn`t anything better. There still isn`t --- and the LaTeX sources from back then can still be worked with and printed with no problem. I still have them.
Provided they only use packages that are compatible with the TeX distro you've got available.
Even if they don`t, I can still edit them and make adjustments if necessary. If anything fails, I can print the sources as they are. Had I used some WYSIWYG word processing software instead, I wouldn`t even be able to read them anymore because this word processing software doesn`t exist anymore since a long time.
I don`t know what would be better than packages. I wish there were more packages and am only saying it`s too difficult to make them. Perhaps there`s good reason for it, perhaps it can be made easier. Someone who knows how to make them might be able to tell.
Ultimately writing a basic spec file is pretty simple, if you can do the configure-make-install cycle it should be straightforward enough.
Is there some documentation about how one would make a package that can be integrated into Fedora?
How would I find out what the package would have to depend on?
On 03/23/2014 04:41 PM, Chris Murphy wrote:
On Mar 23, 2014, at 2:36 AM, Ralf Corsepius rc040203@freenet.de wrote:
On 03/23/2014 08:56 AM, Frank Murphy wrote:
On Sat, 22 Mar 2014 18:29:02 +0000 Timothy Murphy gayleard@eircom.net wrote:
Ed Greshko wrote:
Does anybody nowadays actually burn CDs or DVDs?
Yes, because unlike USB-sticks, I am archiving them and resort to using them as "ready-to-use/off-the-bookshelf" media, e.g. in cases of "emergencies".
Offtopic but for what it's worth pretty much all dye based DVD is effectively disqualified from archiving because of the massive variance in errors, with the initial error rates (immediately after writing) busting the relevant standard (ISO/IEC29121). To even consider it requires ISO/IEC 10995 discs, a dedicated archival recorder, periodic testing strategy to know when to migrate data to new discs, and proper storage conditions (low temperature and humidity).
Sure, but I am not talking about "archiving for eternity".
I am talking about archiving for a couple of years (Typically life-time+ of a Linux-distro) to have a last-resort fallback at hand in situations, when no free usb-stick, no network or functional machine is at hand. These situations are rare, nevertheless they occasionally happen.
In short, I adopted the habit to carry a Fedora-DVD in my Fedora-notebooks' case, because this had saved me from more serious troubles several times. I certainly could carry a USB-stick, instead ;)
Ralf
On 23 March 2014 23:08, lee lee@yun.yagibdah.de wrote:
Ian Malone ibmalone@gmail.com writes:
There is no-one on the planet who thinks emacs is minimalistic
I`m not so sure about that. There are even ppl who have never heard about emacs, and if they`d see it, they might very well think it is.
"Eight Megs and Constantly Swapping" This is a program which has the reputation of having a module for everything. Why are people who have never heard of it relevant? http://xkcd.com/378/ Perhaps even more pertinently: http://xkcd.com/1172/
wasn`t anything better. There still isn`t --- and the LaTeX sources from back then can still be worked with and printed with no problem. I still have them.
Provided they only use packages that are compatible with the TeX distro you've got available.
Even if they don`t, I can still edit them and make adjustments if necessary. If anything fails, I can print the sources as they are. Had I used some WYSIWYG word processing software instead, I wouldn`t even be able to read them anymore because this word processing software doesn`t exist anymore since a long time.
Yes plain LaTeX is fairly readable in source form, but that's not an absolute requirement for being future proof. Educate yourself on ODF. You could even get the plain text out.
I don`t know what would be better than packages. I wish there were more packages and am only saying it`s too difficult to make them. Perhaps there`s good reason for it, perhaps it can be made easier. Someone who knows how to make them might be able to tell.
Ultimately writing a basic spec file is pretty simple, if you can do the configure-make-install cycle it should be straightforward enough.
Is there some documentation about how one would make a package that can be integrated into Fedora?
How would I find out what the package would have to depend on?
There is, but I'm finished googling stuff for you.
On 03/23/2014 07:12 PM, lee wrote:
Joe Zeff joe@zeff.us writes:
On 03/23/2014 12:34 PM, lee wrote:
I`m somewhat surprised that the feeling of apparent desinterest of the makers of Fedora in what its users think seems kinda widespread under its users. Perhaps it`s a wrong impression; if not, it may be something for Fedora.next to address.
Well, Fedora is Gnome-centric, and many of us get the same impression about the Gnome devs.
Well, I don`t use gnome, though I understand what you mean. Still when makers of software X appear desinterested, that doesn`t mean that makers of software Y need to or should appear the same way.
Besides, Fedora.next shouldn`t be limited to gnome ...
I would just like to say - That I use the current version of Gnome ,and coming from a strictly Windows shop at work? I like it. I have tried XFCE, Cinnamon, Enlightenment, IceWM, MATE, KDE, LXDE, FluxBox, and a few others, and for ME? Gnome is fine. I don't have problems with it or issue. I'm not running the latest & greatest in regards to hardware, I'm "middle-of-the-road" and everything works. I for one hope they never stop the work on Gnome, because it suits me just fine...
EGO II
Ian Malone ibmalone@gmail.com writes:
On 23 March 2014 23:08, lee lee@yun.yagibdah.de wrote:
Ian Malone ibmalone@gmail.com writes:
There is no-one on the planet who thinks emacs is minimalistic
I`m not so sure about that. There are even ppl who have never heard about emacs, and if they`d see it, they might very well think it is.
"Eight Megs and Constantly Swapping" This is a program which has the reputation of having a module for everything. Why are people who have never heard of it relevant?
Because their existence disproves that nobody thinks that emacs is minimalistic.
Even if they don`t, I can still edit them and make adjustments if necessary. If anything fails, I can print the sources as they are. Had I used some WYSIWYG word processing software instead, I wouldn`t even be able to read them anymore because this word processing software doesn`t exist anymore since a long time.
Yes plain LaTeX is fairly readable in source form, but that's not an absolute requirement for being future proof. Educate yourself on ODF. You could even get the plain text out.
ODF didn`t exist twenty years ago.
On 24 March 2014 10:09, lee lee@yun.yagibdah.de wrote:
Ian Malone ibmalone@gmail.com writes:
On 23 March 2014 23:08, lee lee@yun.yagibdah.de wrote:
Ian Malone ibmalone@gmail.com writes:
There is no-one on the planet who thinks emacs is minimalistic
I`m not so sure about that. There are even ppl who have never heard about emacs, and if they`d see it, they might very well think it is.
"Eight Megs and Constantly Swapping" This is a program which has the reputation of having a module for everything. Why are people who have never heard of it relevant?
Because their existence disproves that nobody thinks that emacs is minimalistic.
That's quite clearly a logical fallacy, but I don't think there's anything productive left to say here.
On Sun, Mar 23, 2014 at 10:56:13PM +0100, lee wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
I'm sorry but the installer denying /usr on its own partition on F17 is the right thing to do. I believe F17 introduced something called usr-move, meaning all the binaries in /bin /sbin are actually hardlinks/symlinks to /usr/bin and /usr/sbin. I believe this was a multi-distribution effort. In such a configuration, there is no justification or gain of putting it in a separate partition, on top of that the booting process becomes quite complicated.
On Sun, Mar 23, 2014 at 08:34:55PM +0100, lee wrote:
I`m somewhat surprised that the feeling of apparent desinterest of the makers of Fedora in what its users think seems kinda widespread under its users. Perhaps it`s a wrong impression; if not, it may be something for Fedora.next to address.
I think it's unfair impression. Some of the developers only care about their own area, and perhaps the users of that thing but not the project overall, sure. But overall, we care very much.
The catch is: it's hard to get a good sense of what the userbase as a whole thinks. It's easy to conflate that with "care about what some users are repeating very loudly on a mailing list". That's certainly _some_ input, but it's a skewed view of the actual world.
Many of the better possible ways to measure are very expensive, but overall, there is one which is straightforward and effective: users who represent a large enough community will have some percentage of people willing and able to contribute by becoming Fedora developers, and those people guide the project through their actions. That's why we have Gnome, KDE, LXDE, and MATE-Compiz desktop spins -- and, pointedly, not a fvwm one. If you really think that this is the best course for Fedora, I encourage you to step up and create one. (See https://fedoraproject.org/wiki/Spins_Process.)
Or, if that's not really you're thing, you could step back and focus on what you are suggesting is a bigger problem -- getting user input into Fedora. How could that be done better? Surveys? More user testing? An active "User Feedback SIG"?
Allegedly, on or about 23 March 2014, lee sent:
I`m somewhat surprised that the feeling of apparent desinterest of the makers of Fedora in what its users think seems kinda widespread under its users. Perhaps it`s a wrong impression; if not, it may be something for Fedora.next to address.
Many users probably don`t care as long as Fedora works for them --- and when it doesn`t anymore, they just pick something else?
I reckon it's the case for most OSs that /most/ users don't really care much about what they're using, nor how it works. The large number of clueless people using computers would seem to be evidence of that. It's a lesser number of people that have concerns about their OS. Maybe that's why developers are less concerned about public opinions.
Matthew Miller mattdm@fedoraproject.org writes:
On Sun, Mar 23, 2014 at 08:34:55PM +0100, lee wrote:
I`m somewhat surprised that the feeling of apparent desinterest of the makers of Fedora in what its users think seems kinda widespread under its users. Perhaps it`s a wrong impression; if not, it may be something for Fedora.next to address.
I think it's unfair impression. Some of the developers only care about their own area, and perhaps the users of that thing but not the project overall, sure. But overall, we care very much.
... and time and resources are limited.
The catch is: it's hard to get a good sense of what the userbase as a whole thinks. It's easy to conflate that with "care about what some users are repeating very loudly on a mailing list". That's certainly _some_ input, but it's a skewed view of the actual world.
Many of the better possible ways to measure are very expensive, but overall, there is one which is straightforward and effective: users who represent a large enough community will have some percentage of people willing and able to contribute by becoming Fedora developers, and those people guide the project through their actions.
Are users making packages substantially different from users giving input through a mailing list? In both cases, you have a number of users, and some of them do something, i. e. are active on the mailing list or are active by making packages.
The ones making packages probably have more influence. Is it supposed to be like that?
That's why we have Gnome, KDE, LXDE, and MATE-Compiz desktop spins -- and, pointedly, not a fvwm one. If you really think that this is the best course for Fedora, I encourage you to step up and create one. (See https://fedoraproject.org/wiki/Spins_Process.)
I don`t understand why every possible choice should require making a distribution on it`s own.
Or, if that's not really you're thing, you could step back and focus on what you are suggesting is a bigger problem -- getting user input into Fedora. How could that be done better? Surveys? More user testing? An active "User Feedback SIG"?
I think that a mailing list like this one can provide a lot of input and that extracting it can be a problem. It`s not like all package maintainers could read all posts here in order to figure out what users might want.
Surveys might be a good idea, though someone would have to put them together, and someone would have to extract the information from them. Both of that isn`t easily done, either.
How do you currently find out what users want?
Suvayu Ali fatkasuvayu+linux@gmail.com writes:
On Sun, Mar 23, 2014 at 10:56:13PM +0100, lee wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
I'm sorry but the installer denying /usr on its own partition on F17 is the right thing to do. I believe F17 introduced something called usr-move, meaning all the binaries in /bin /sbin are actually hardlinks/symlinks to /usr/bin and /usr/sbin. I believe this was a multi-distribution effort. In such a configuration, there is no justification or gain of putting it in a separate partition, on top of that the booting process becomes quite complicated.
/usr belongs on it`s own partition. And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
On Mar 23, 2014, at 4:18 PM, lee lee@yun.yagibdah.de wrote:
"Powell, Michael" Michael_Powell@mentor.com writes:
Then they need to turn around at once! Leaving users in the dark about what`s going on makes things more difficult for them. Taking away choices limits the use of the software to the point where it eventually becomes unusable.
Good design, clarity, good documentation and letting the user know what choices there are are required when you want to achieve ease of use.
I don't think the installer is as bad as most people on this list are complaining about; it's certainly not leaving any users in the dark.
It doesn`t give you choices. It leaves you in the dark about that it is somehow possible to use an non-gui installer and to do a minimal install. It leaves you in the dark about what exactly happens when you do the partitioning and with trying to figure out how get the partitioning you want.
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
Please post the bugzilla URL.
Most of the size reporting problems like this are non-contiguous sections of free space being added up and reported as Available space; but the request is for a partition size greater than the largest contiguously available space.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
It isn't going to get better complaining about it on this list. Do you have bugzilla IDs, and if so post them. If not, then how do you expect the behavior to get any better? Magic?
A large majority of the information from the older installer is still there, it's just up to the user to seek it out.
I don`t know anything about "the older installer" or how to seek out information about it; I didn`t even know that there is an "older installer". The first Fedora installer I used was the one with F17.
Fedora 17 is the old installer. Fedora 18 begins the new installer.
I don`t know what you mean. The Debian installer got more options and some more clarity which was an improvement. Otherwise it didn`t change, you just do country and keyboard setup --- which is missing in Fedoras installer, there was no way to tell it that I have a German keyboard --- network setup if you don`t use DHCP, partitioning, a bit of package selection if you want to, and then it installs.
Fedora 18, 19 and 20 have a keyboard spoke in the installer which is how you tell it you want to use a German keyboard layout.
It`s easy and straightforward as it used to be for the last twenty years, and I never had trouble using it. Why suddenly make installing such a PITA like Fedoras installer does?
Please don't ask silly questions that propose the intended design goal was to piss users off, it's irritating.
Chris Murphy
On 24 March 2014 12:45, lee lee@yun.yagibdah.de wrote:
/usr belongs on it`s own partition. And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
I think that's a very 1980s, or early-1990s, way of looking at it.
Since the normal way to boot a PC now is a complete functioning OS on a single removable-media volume - be that an optical disk or USB flash media - most of the rationale for splitting up the bits of the /usr tree have long ceased to apply. The smallest hard disks available today (~500GB) are roughly 2 orders of magnitude bigger than is needed for a full Linux desktop install (~5GB). It is not possible to buy a new computer without a graphical display.
There is no need for separating out admin binaries, user binaries, local binaries, graphical binaries etc. any more, and hasn't been for about 2 decades.
I think it's a brilliant, if brave, idea of Fedora to get rid of a historical distinction that is now pointless, but it's planned and discussed and decided, as far as I know:
https://fedoraproject.org/wiki/Features/UsrMove
http://www.freedesktop.org/wiki/Software/systemd/TheCaseForTheUsrMerge/
On 23 March 2014 21:56, lee lee@yun.yagibdah.de wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
As I have commented elsewhere, I think this is a 1980s style of thinking. Things have changed. Move on. Sorry, but they have; see the links elsewhere.
RAID isn`t exotic, either. Disks do fail, the only question is when, and I neither want to lose data, nor the hassle.
Sure, I use it on all my servers.
Installing on a laptop requires encrypted partitions. They can be stolen too easily.
I have never ever used this and never expect or plan to. I suggest that your blanket statement is too sweeping.
I never separate out /tmp or /var or /usr/local - I only ever use / and /home basically.
I always use separate partitions. It has lots of advantages.
As I said, I use / and /home and advise against combining them.
Personally I think that's enough. I am not disputing your reasons, but AIUI, Fedora is trying out a move to flatten and simplify the way-too-complex directory hierarchy. It's happened. The decision is made. Deal with it, move on.
/var can get full, and it`s written to, same goes for /tmp.
As I said elsewhere: when the smallest new HD you can buy is half a terabyte (and even SSDs start at half that) this really isn't a big issue any more.
How do you mount /usr read-only?
I don't. Never have in 26y of Unix systems support. For rescue, now, I boot off a LiveDVD or LiveUSB.
Especially on a server, it`s a good idea to mount everything read-only that you can. When you have several disks, you can do your partitioning in such a way that you get better performance.
An illustrative anecdote, for what it's worth:
Back in 1995 I was the testing labs manager for a leading UK computer magazine. In an article, I recommended partitioning 1.2GB disks with a dedicated swap partition at the end. This meant the main partition could be under 1GB and thus use far more space-efficient 8kB blocks. (This is on Win95 or NT 3.5, before FAT32 was invented.)
A system manufacturer complained that putting swap on the end of the drive would kill performance. I disagreed. He complained more.
So he came in and I set up a test and showed him that, to 2 decimal places in a percentage-based benchmark score, i.e. well below measurement error, there was absolutely *no* difference between the speeds of different areas of the disk.
This is in 1995, when LBA addressing was new, 1.2GB was a big disk, and DMA was just starting to appear for EIDE drives.
Back then, the difference was not measurable.
That was nearly 20y ago.
Now, it is not real. It is not there any more. Disks are a thousand times bigger and faster now. This stuff does not matter any more and hasn't since before Linux 2.0 was released.
Especially when you have a server, you may need a (pretty much) granted capacity on /var or /tmp to make sure it will continue to operate --- without separate partitions, your users may fill up the disks ...
Users shouldn't be able to write stuff to / at all! Only to /home or below, or dedicated data partitions. Where /usr or /var is should not matter to them.
Nowadays you may have SSDs which supposedly last longer when not written much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
Machines come with dozens of gigs of RAM now. I'm not sure there's much argument for swap at all, and personally, I use tmpfs for better performance and a self-cleaning /tmp tree.
Why wouldn`t you use different partitions? I can see it (and have done it) for when the available disk capacity is extremely limited, but otherwise it doesn`t make any sense and has nothing but disadvantages.
Exactly. One splits stuff up when space is an issue. When it isn't, one doesn't need to.
It`s merely a reasonable standard thing to use separate partitions and a requirement to use RAID, and encrypted partitions for laptops, not something in any way unusual. Of course I expect an installer to handle that as well as using a single, unencrypted partition on a single disk.
Actually, I agree, but there must be /some/ limits to the granularity!
And it`s not too difficult. The installer doesn`t need to do the partitioning, the user does it. The installer only needs to give the user a good tool to do the partitioning the user wants and let them use it. Good tools to do partitioning are already available, and the installer doesn`t need to re-invent the wheel in that.
I mostly agree, but bear in mind that the installer must cope with both experts and novices. That's a tough call.
I'd say it's /too/ simplified at the moment, though, and I think you might agree...?
Perhaps it even shouldn`t. Why force the user to learn how to use yet another partitioning tool they even rarely use unless they install Fedora all the time? Why not give them a choice, like either cfdisk or parted, then tell the installer what to do with each partition and let them switch between these until they are done --- or let the installer do whatever partitioning it wants, which means that all existing data on the disks will be deleted.
Mostly, I would agree, actually.
On 24 March 2014 01:33, Patrick O'Callaghan pocallaghan@gmail.com wrote:
On Sun, 2014-03-23 at 18:24 +0000, Liam Proven wrote:
On 22 March 2014 18:25, Ed Greshko ed.greshko@greshko.com wrote:
Make sure you file a bugzilla.
How/why? It's not a bug.
If it doesn't work as it's supposed to, it's a bug.
An intentional design feature that causes a problem in an unexpected or unconsidered use-case is not what I'd call a bug, but hey, have it your own way. :-)
Talking about it on this list will not result in any action.
Won't it? Why not? Are there multiple non-overlapping user communities for Fedora, then? A problem is a problem; where it is reported shouldn't matter, should it?
Seriously? Yes, there are multiple partially-overlapping communities.
Yes seriously. I'm new here. I have no idea who does or does not read this.
Many of the developers don't read this list. Many of the people on this list don't read the Test list. If you want to document a bug then use Bugzilla.
As I have said, I don't think that I /can/ document it well enough to be of any use.
On Mon, 24 Mar 2014, lee wrote:
Bill Oliver vendor@billoblog.com writes:
On Sun, 23 Mar 2014, lee wrote:
[snip] If she had known that she didn`t need to throw in this or that, wouldn`t that be easier for everyone?
Not necessarily. Sometimes neither that absent-minded aunt nor the avaricious nephew know what is most pleasing until it's taken out and played with.
That can be an important thing to do. It requires listening to the users and informing and preparing them beforehand, and you have to have something available to fall back to, or a fix, in case the new toy doesn`t work out. If you leave them suddenly stranded like Debian did, they will leave you behind.
Much of what we consider very important today was considered stupid when it first came out.
Like?
Well, the automobile, aircraft, and personal computer immediately come to mind.
billo
On 23 March 2014 20:19, Chris Murphy lists@colorremedies.com wrote:
On Mar 23, 2014, at 12:24 PM, Liam Proven lproven@gmail.com wrote:
On 22 March 2014 18:25, Ed Greshko ed.greshko@greshko.com wrote:
Make sure you file a bugzilla.
How/why? It's not a bug.
How do you know? You haven't filed a bug report, you haven't presented any logs to anyone to find out if what you're experiencing is in fact intended behavior. You haven't even stated the nature of the installation failure.
I have to say that you are one of the most hostile, confrontational and aggressive people that I've come across in a support community since Richard Morrell of Smoothwall, about 12 years back, who alienated so many users that a bunch of them got together and forked it into IPCop... in large part so that they could have support communities without that man in them, telling them that they were stupid and wrong and should just go away if they didn't like it.
Do you realise this?
Are you doing it deliberately?
It will not. Because this is the user to user forum. To cause a design change in the installer means posting to the Red Hat anaconda-devel@ list. To cause a design change specific to Fedora means posting to the Fedora devel@ list. Bugs, requests for enhancement, and even perceived clearly articulated design flaws should have bug reports filed. Pretty much always when an installer bug is filed, the installer logs should be attached (at least program.log and storage.log).
Well, OK. I am not a developer of or contributor to the product and since I can't even install it on my PC I am not very likely to become one, so I will just have to let it be for now, then.
You're seriously acting like this is your first day at the rodeo. You are supposed to state what did happen,
What, like I did on the 5th of March in the threat entitled "Can't install ƒ20", when I said:
<< Hi there. First post from someone returning to the RH family fold for the first time since about 1996-1999, when I was a RHL user (v4 - v6 full time on my servers, v7/8/9 reviewed for various magazines - then I switched to Caldera, then to SUSE and then to Ubuntu when it came out.)
I'm trying to take a look at Fedora on real hardware as opposed to in a VM. This is on a machine that multi-boots Windows 7, Ubuntu 13.10, Crunchbang 11 and (used to) ElementaryOS.
I am trying to install on /dev/sda. It's a 1TB HD - there's also a 120GB SSD as /dev/sdb with Ubuntu's root filesystem and Win7 on it.
I have a pre-configured set of folders:
/dev/sda10 /home 150GB /dev/sda11 (NTFS shared data) ~750GB /dev/sda5 / 16GB <- this is where I want to put Fedora; it used to be elementary's / /dev/sda6 (Crunchbang /) 16GB /dev/sda7 (FAT32, Windows pagefile) 8GB /dev/sda8 (Linux swap) 8GB
So I need to tell Fedora to put root on sda5 and /home on sda10 and use sda8 for swap. I want the bootloader on sda5 as well. I'm currently using Ubuntu's GRUB, as it's my primary OS, but I plan to replace this with a standalone boot manager.
I managed to get the Fedora installer to format sda5 but then, having 16GB of / + 150GB of /home and 8GB of swap, it said that it didn't have enough space for the 6.7GB of stuff it needed to install and crashed out with a series of Python errors.
Now it can't "see" a distro on sda5, it won't let me choose it as an option. I can't remove and recreate it, either - or at least, I can't see how. I also can't see how to tell it to put the bootloader in the root partition.
Is it me, or is the installer just not flexible enough to cope with this sort of scenario?
including any error messages received
Tricky when the system hangs.
and how this differs from your expectation of what should have happened.
What I expect? Well, in a nutshell, I expect to be able to say:
"Here, >this< is your root, >this< is /home, swap is >there<, put your bootloader in the root FS, now go."
Failing that, to be able to delete an existing root partition, make a new one, say ">that< is your root, go".
Is that so unreasonable?
The installer always keeps logs in /tmp
As I said, it locked up.
- so go try a Fedora baremetal install and when it "won't install" and you're stuck, post the /tmp/anaconda.log, /tmp/program.log, and /tmp/storage.log.
I will have to look up what a "Fedora baremetal install" is. AFAICR I saw no such option in the boot/install image.
You can even fpaste them directly from the shell, just fpaste <filename> and then note the URL and post it.
[Googles "fpaste".]
Ah, the Fedora pastebin. Right. You saved so many characters by not saying that, I entirely understand why you left it to me. </ironic>
I also note that you assume I have working Internet access off the installer. Interesting.
On 03/24/2014 03:12 PM, Liam Proven wrote:
On 24 March 2014 12:45, lee lee@yun.yagibdah.de wrote:
/usr belongs on it`s own partition. And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
I think that's a very 1980s, or early-1990s, way of looking at it.
C'mon, feeling something is oldfashioned is hardly an answer.
Having been able to have /usr on a separate partition was a valuable feature, which now has gone lost. IMNSHO, ruined by naive, inexperienced kids (to use the same tone as you did), who were overwhelmed by the additional complexity supporting this feature had required.
Since the normal way to boot a PC now is a complete functioning OS on a single removable-media volume - be that an optical disk or USB flash media - most of the rationale for splitting up the bits of the /usr tree have long ceased to apply. The smallest hard disks available today (~500GB) are roughly 2 orders of magnitude bigger than is needed for a full Linux desktop install (~5GB).
Wrong. You are forgetting about systems booting from SD-Cards, USB-sticks and other forms of non-volatile memory.
It is not possible to buy a new computer without a graphical display.
Wrong. Most servers typically are headless, and if they have a graphic card-build-in, it's usually inaccessible or unused.
There is no need for separating out admin binaries, user binaries, local binaries, graphical binaries etc. any more, and hasn't been for about 2 decades.
Right, there is no strong necessity, nevertheless having these still would make sense.
I think it's a brilliant, if brave, idea of Fedora to get rid of a historical distinction that is now pointless, but it's planned and discussed and decided, as far as I know:
I disagree. IMNSHO, UsrMove was a prominent epic fail in the long serious faulty decisions Fedora's leadership has committed.
Ralf
On Mon, Mar 24, 2014 at 02:47:18PM +0000, Liam Proven wrote:
I have to say that you are one of the most hostile, confrontational
I said this before and I'll say it again if I have to (but I hope I don't). Please, everyone, check out the Code of Conduct. It's linked in the footer of every post, but here's a reminder:
https://fedoraproject.org/code-of-conduct.
Let's all work on elevating the level of discourse without getting into who is or isn't starting it. On the topic at hand, remember that a lot of people have put significant work into making the installer interface intuitive, easy to use, powerful, friendly, and all sorts of other things all at once. Many of the comments going by seem very dismissive of that, and it's easy to see how that escalates into hot tempers.
It's one thing to say "For me, this has missed the mark in this way", and another to call all the work terrible or to make statements which imply actual antagonism. I don't mean that it's perfect (it's not), or that the developers are perfect (they're human!), or that the frustrations aren't real -- just please be aware how that might feel to people on the other side of the equation.
Please, everyone, assume good faith on the part of the other. Software developers have the best intentions in making changes, and frustrated users are just frustrated.
On 24 March 2014 15:02, Ralf Corsepius rc040203@freenet.de wrote:
On 03/24/2014 03:12 PM, Liam Proven wrote:
On 24 March 2014 12:45, lee lee@yun.yagibdah.de wrote:
/usr belongs on it`s own partition. And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
I think that's a very 1980s, or early-1990s, way of looking at it.
C'mon, feeling something is oldfashioned is hardly an answer.
Well, actually, sometimes, yes, it is. Same as the decision to drop support for i386 from the kernel, or the fact that no installers default to ext2 any more.
Having been able to have /usr on a separate partition was a valuable feature, which now has gone lost. IMNSHO, ruined by naive, inexperienced kids (to use the same tone as you did), who were overwhelmed by the additional complexity supporting this feature had required.
I am not saying you're wrong, merely that I personally haven't seen a use or need for it since about 1989 and I found the reasoning for its collapse and merger to be sound.
Wrong. You are forgetting about systems booting from SD-Cards, USB-sticks and other forms of non-volatile memory.
Is Fedora a suitable OS for such hardware? I'd argue not, myself.
Wrong. Most servers typically are headless, and if they have a graphic card-build-in, it's usually inaccessible or unused.
I am actually an IT professional - no, honestly, really I am - and every single rackmount server I've used in the last few years still has an SVGA port on it.
Right, there is no strong necessity, nevertheless having these still would make sense.
There was an argument; it was decided not. I wasn't involved. I happen to agree, but I can't change it, so there's no point telling me! :¬)
I disagree.
Yeah, I guessed. :¬)
IMNSHO, UsrMove was a prominent epic fail in the long serious faulty decisions Fedora's leadership has committed.
We-eeeelll... I am not sure that I could overall disagree with the general thrust of your argument there. :¬)
On 24 March 2014 15:08, Matthew Miller mattdm@fedoraproject.org wrote:
On Mon, Mar 24, 2014 at 02:47:18PM +0000, Liam Proven wrote:
I have to say that you are one of the most hostile, confrontational
I said this before and I'll say it again if I have to (but I hope I don't). Please, everyone, check out the Code of Conduct. It's linked in the footer of every post, but here's a reminder:
https://fedoraproject.org/code-of-conduct.
Let's all work on elevating the level of discourse without getting into who is or isn't starting it. On the topic at hand, remember that a lot of people have put significant work into making the installer interface intuitive, easy to use, powerful, friendly, and all sorts of other things all at once. Many of the comments going by seem very dismissive of that, and it's easy to see how that escalates into hot tempers.
It's one thing to say "For me, this has missed the mark in this way", and another to call all the work terrible or to make statements which imply actual antagonism. I don't mean that it's perfect (it's not), or that the developers are perfect (they're human!), or that the frustrations aren't real -- just please be aware how that might feel to people on the other side of the equation.
Please, everyone, assume good faith on the part of the other. Software developers have the best intentions in making changes, and frustrated users are just frustrated.
Good point, well made. I hope that my observations did not seem like a personal attack. They were not meant to be.
On Mar 23, 2014, at 3:56 PM, lee lee@yun.yagibdah.de wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
Old news.
https://fedoraproject.org/wiki/Features/UsrMove#I_have_.2Fusr_as_a_separate_...
and
http://www.freedesktop.org/wiki/Software/systemd/separate-usr-is-broken/
RAID isn`t exotic, either.
It kinda is. The definition of exotic is "not ordinarily encountered." And that's even if you look at just the Linux universe, because overwhelmingly most users don't use it. If you look at the rest of the computing world it's either not an install time option or not possible.
I always use separate partitions. It has lots of advantages.
It also has many disadvantages. Hence LVM thinp and btrfs subvolumes as alternatives.
I might split off /var on a server but I'd need a remarkably persuasive use case, and on servers, I use extra-stable distros without GUIs, not something like Fedora.
/var can get full, and it`s written to, same goes for /tmp.
Use quotas.
Nowadays you may have SSDs which supposedly last longer when not written much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
It's in the realm of 20+GB written per day every day, for the warranty period. If you're doing that, get an enterprise SSD. Or stick with HDDs.
Why wouldn`t you use different partitions? I can see it (and have done it) for when the available disk capacity is extremely limited, but otherwise it doesn`t make any sense and has nothing but disadvantages.
It's cute when people project their world view in such narrow terms. Looking over at Windows and OS X, for 20+ years they've had at most two partitions, one for boot and one for everything else. OS X only just went from one partition for everything to two just a few releases ago in order to support full disk encryption (around a decade after Linux had it, but then it's also on-the-fly COW online convertible bidirectionally). So there isn't an inherent good for partitioning. It's useful for certain use cases. It's a negative for others.
But this just illustrates the breadth of scenarios a successful installer must cope with!
It`s merely a reasonable standard thing to use separate partitions and a requirement to use RAID, and encrypted partitions for laptops, not something in any way unusual.
It is in fact unusual. What you're doing is proposing that what works for you, as a default for everyone. And your arguments for changing the paradigm are completely uncompelling.
"a requirement to use RAID" is particularly irritating because you're saying everyone without boot from SAN capability, or a laptop, should be required to have two like sized drives to do an install. Otherwise it wouldn't be "required". So what you're writing doesn't even make sense.
Of course I expect an installer to handle that as well as using a single, unencrypted partition on a single disk.
And it`s not too difficult. The installer doesn`t need to do the partitioning, the user does it. The installer only needs to give the user a good tool to do the partitioning the user wants and let them use it. Good tools to do partitioning are already available, and the installer doesn`t need to re-invent the wheel in that.
Perhaps it even shouldn`t. Why force the user to learn how to use yet another partitioning tool they even rarely use unless they install Fedora all the time?
The Fedora installer will use existing layouts: partitions, PV/VGs, Btrfs, and md RAID. Click on the existing logical device and then specify a mount point for it.
Why not give them a choice, like either cfdisk or parted, then tell the installer what to do with each partition and let them switch between these until they are done
Use kickstart?
I think letting users switch between two tools means both tools are failures, or the user is neurotic. Pick a tool for a task, move on to the next tool. Switching back and forth isn't going to be stable. You're talking about a tiny percent of users doing what you suggest, which means an infinitesimal number are testing it.
Seldom used or seldom tested things have no place in a GUI installer.
--- or let the installer do whatever partitioning it wants, which means that all existing data on the disks will be deleted.
This is the most commonly supported install/reinstall use case. It is highly stable, highly successful, pretty much never fails, takes far far less development and test time, and has little potential for regressions.
Chris Murphy
On Mar 24, 2014, at 6:27 AM, Suvayu Ali fatkasuvayu+linux@gmail.com wrote:
On Sun, Mar 23, 2014 at 10:56:13PM +0100, lee wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
I'm sorry but the installer denying /usr on its own partition on F17 is the right thing to do. I believe F17 introduced something called usr-move, meaning all the binaries in /bin /sbin are actually hardlinks/symlinks to /usr/bin and /usr/sbin. I believe this was a multi-distribution effort. In such a configuration, there is no justification or gain of putting it in a separate partition, on top of that the booting process becomes quite complicated.
That's all correct.
Chris Murphy
On Mon, Mar 24, 2014 at 2:12 PM, Liam Proven lproven@gmail.com wrote:
Since the normal way to boot a PC now is a complete functioning OS on a single removable-media volume - be that an optical disk or USB flash media
Uhhh... wow. That's quite some selection bias you have going on there. For the record, I'm not aware of a single Linux user that boots their system that way. To hear it described as "normal" is somewhat strange.
Tet
On 24 March 2014 15:32, Tethys tethys@gmail.com wrote:
Since the normal way to boot a PC now is a complete functioning OS on a single removable-media volume - be that an optical disk or USB flash media
Uhhh... wow. That's quite some selection bias you have going on there. For the record, I'm not aware of a single Linux user that boots their system that way. To hear it described as "normal" is somewhat strange.
Perhaps I should have been clearer; my apologies. I meant for installation, recovery or repair purposes.
I am not aware of /any/ modern distro that has boot, root & source media any more, or which supports installation from multiple floppies. Even offering installation from multiple CDs is rare to unheard-of now. Indeed CD support itself is disappearing - they're too small for OSes here in the 2nd decade of C21.
On Mar 24, 2014, at 6:45 AM, lee lee@yun.yagibdah.de wrote:
Suvayu Ali fatkasuvayu+linux@gmail.com writes:
On Sun, Mar 23, 2014 at 10:56:13PM +0100, lee wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
I'm sorry but the installer denying /usr on its own partition on F17 is the right thing to do. I believe F17 introduced something called usr-move, meaning all the binaries in /bin /sbin are actually hardlinks/symlinks to /usr/bin and /usr/sbin. I believe this was a multi-distribution effort. In such a configuration, there is no justification or gain of putting it in a separate partition, on top of that the booting process becomes quite complicated.
/usr belongs on it`s own partition.
As if no one has ever said that before, and as if it convinced even one thinking person to change their mind.
Fedora has never defaulted to separate /usr partition. It's been two years since this was decided. That you're still experiencing cognitive dissonance over this ancient long ago resolve topic is your problem, not anyone else's.
And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
bin lib lib64 are symlinks to their locations in /usr.
Chris Murphy
On 03/24/2014 07:12 AM, Liam Proven wrote:
Since the normal way to boot a PC now is a complete functioning OS on a single removable-media volume - be that an optical disk or USB flash media
Where did you ever get that idea? Very very few users make a habit of booting from a LiveCD or USB, which is what you're describing.
On 03/24/2014 04:28 PM, Chris Murphy wrote:
On Mar 23, 2014, at 3:56 PM, lee lee@yun.yagibdah.de wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
Old news.
https://fedoraproject.org/wiki/Features/UsrMove#I_have_.2Fusr_as_a_separate_...
and
http://www.freedesktop.org/wiki/Software/systemd/separate-usr-is-broken
Correct, /usr on a separate partition had been broken in Red Hat based distros for a long time. Why? Because maintainers did not take care about it.
Ralf
On Mar 24, 2014, at 8:36 AM, Liam Proven lproven@gmail.com wrote:
On 24 March 2014 01:33, Patrick O'Callaghan pocallaghan@gmail.com wrote:
On Sun, 2014-03-23 at 18:24 +0000, Liam Proven wrote:
On 22 March 2014 18:25, Ed Greshko ed.greshko@greshko.com wrote:
Make sure you file a bugzilla.
How/why? It's not a bug.
If it doesn't work as it's supposed to, it's a bug.
An intentional design feature that causes a problem in an unexpected or unconsidered use-case is not what I'd call a bug, but hey, have it your own way. :-)
Right, so in your view ten years of install failures to baremetal is an intentional feature?
Talking about it on this list will not result in any action.
Won't it? Why not? Are there multiple non-overlapping user communities for Fedora, then? A problem is a problem; where it is reported shouldn't matter, should it?
Seriously? Yes, there are multiple partially-overlapping communities.
Yes seriously. I'm new here. I have no idea who does or does not read this.
How did you find out about the list and how did you sign up for it? Doing several Google searches for Fedora user lists, the top most result is this page which makes it abundantly clear there are many Fedora lists.
https://lists.fedoraproject.org/mailman/listinfo
Many of the developers don't read this list. Many of the people on this list don't read the Test list. If you want to document a bug then use Bugzilla.
As I have said, I don't think that I /can/ document it well enough to be of any use.
Oh so that's a reason to not even try? As I have said, you can document it. I've told you where to find the logs, and more than one way to make them available. Your lack of curiosity and active resistance to finding out the cause of these install failures is suspicious.
Chris Murphy
On 03/24/2014 10:38 AM, Chris Murphy wrote:
On Mar 24, 2014, at 6:45 AM, lee lee@yun.yagibdah.de wrote:
Suvayu Ali fatkasuvayu+linux@gmail.com writes:
On Sun, Mar 23, 2014 at 10:56:13PM +0100, lee wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
I'm sorry but the installer denying /usr on its own partition on F17 is the right thing to do. I believe F17 introduced something called usr-move, meaning all the binaries in /bin /sbin are actually hardlinks/symlinks to /usr/bin and /usr/sbin. I believe this was a multi-distribution effort. In such a configuration, there is no justification or gain of putting it in a separate partition, on top of that the booting process becomes quite complicated.
/usr belongs on it`s own partition.
As if no one has ever said that before, and as if it convinced even one thinking person to change their mind.
Fedora has never defaulted to separate /usr partition. It's been two years since this was decided. That you're still experiencing cognitive dissonance over this ancient long ago resolve topic is your problem, not anyone else's.
And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
bin lib lib64 are symlinks to their locations in /usr.
Chris Murphy
Just to add to the topic, the reason we once had separate /usr, /var, etc. partitions was to make it easier to recover filesystems. It took a long time for fsck to search partitions when the drives were small and then to recover. With todays drives being the sizes they are, it still makes sense (in my mind) to keep the partitions separate for the same reason. Just because systems have gotten faster it's still time consuming to have to try to recover data on huge partitions...separation mitigates that issue to a great degree. Another point, there was a comment in the thread made that admin binaries, user binaries, application binaries don't need to be kept separate anymore...I call B.S. on that. Part of the reason that /usr, /bin, /lib, etc. were partitioned as they were was to allow that happen (both from a recovery perspective and, frankly, from a security perspective...kind of the start of the concept of sandboxing). I see no argument to be made to *allow them to be together except laziness. I, for one and probably a minority, try to keep as much of the partitioning "as it was in the '80s" as I can for those same reasons...does it mean that up-front I have to do a little more work? Sure...do I reap benefits from it? Absolutely! (as you can probably tell I was not a proponent of the usr merge that happened a few years ago).
Kevin
On 24 March 2014 15:51, Chris Murphy lists@colorremedies.com wrote:
Right, so in your view ten years of install failures to baremetal is an intentional feature?
And again with the challenging and hostility.
I have not tried every single release; I tend to do distro roundups maybe every 3-5 years.
The particular problem with F20 is one I've never seen before, in Fedora or in any other OS ever.
How did you find out about the list and how did you sign up for it? Doing several Google searches for Fedora user lists, the top most result is this page which makes it abundantly clear there are many Fedora lists.
Uhuh. And from this I am meant to *just know* by my powerful psychic powers who reads it or not, am I?
Oh so that's a reason to not even try?
Actually, yeah.
As I have said, you can document it. I've told you where to find the logs, and more than one way to make them available. Your lack of curiosity and active resistance to finding out the cause of these install failures is suspicious.
No, I think *you're* suspicious, and that is a rather different thing.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 03/24/2014 09:22 AM, lee wrote:
Matthew Miller mattdm@fedoraproject.org writes:
On Sun, Mar 23, 2014 at 08:34:55PM +0100, lee wrote:
I`m somewhat surprised that the feeling of apparent desinterest of the makers of Fedora in what its users think seems kinda widespread under its users. Perhaps it`s a wrong impression; if not, it may be something for Fedora.next to address.
I think it's unfair impression. Some of the developers only care about their own area, and perhaps the users of that thing but not the project overall, sure. But overall, we care very much.
... and time and resources are limited.
The catch is: it's hard to get a good sense of what the userbase as a whole thinks. It's easy to conflate that with "care about what some users are repeating very loudly on a mailing list". That's certainly _some_ input, but it's a skewed view of the actual world.
Many of the better possible ways to measure are very expensive, but overall, there is one which is straightforward and effective: users who represent a large enough community will have some percentage of people willing and able to contribute by becoming Fedora developers, and those people guide the project through their actions.
Are users making packages substantially different from users giving input through a mailing list? In both cases, you have a number of users, and some of them do something, i. e. are active on the mailing list or are active by making packages.
The ones making packages probably have more influence. Is it supposed to be like that?
Frankly, yes. Feedback on a list is fine, but anyone can say "Hey, I wish it was more like this:", but ultimately it will be up to someone to actually implement that change. The ones who go and do the work are the ones who have the final say on what happens.
If something is interesting or important enough to an individual, then they can decide to either implement it themselves or seek out the specific resources to do so for them (either by hiring a contractor or talking someone with the requisite skills into doing it for them).
Very little change or improvement ever happens because a lot of people talked about doing something for years. Things change because someone actually goes and makes it happen. That's the culture we try to encourage in Fedora. Sometimes it means that a very dedicated group of individuals goes and implements something that a lot of people disagree with. When enough of those people disagree, someone will probably come forth and try to either fix or replace that subsystem. It happens with desktop environments, it happens with low-level system components and it's always going to be that way.
That's why we have Gnome, KDE, LXDE, and MATE-Compiz desktop spins -- and, pointedly, not a fvwm one. If you really think that this is the best course for Fedora, I encourage you to step up and create one. (See https://fedoraproject.org/wiki/Spins_Process.)
I don`t understand why every possible choice should require making a distribution on it`s own.
You're confusing a spin with a distribution (and a product with a distribution).
Both products and spins are curated sets of packages from a single distribution (Fedora). Each one has its own reason for existing (in the case of the desktop spins, it's basically to show off a particular piece of technology).
For the Products, we're working to establish specific *solutions*. Recognizing that most people install an operating system so that, well, they can operate their system, we're trying to build solutions for three common use-cases so that newcomers to the Fedora Project don't feel like they need to make a thousand individual package choices to get their system running. There will always be people who want to do that, and we'll continue to cater to them by having the wider package set remain available (as well as the spins process so people who care enough can build new install-and-deployment media).
Or, if that's not really you're thing, you could step back and focus on what you are suggesting is a bigger problem -- getting user input into Fedora. How could that be done better? Surveys? More user testing? An active "User Feedback SIG"?
I think that a mailing list like this one can provide a lot of input and that extracting it can be a problem. It`s not like all package maintainers could read all posts here in order to figure out what users might want.
Mailing lists are a *start*, but you also have to recognize that you're dealing with a self-selected set of responders. Specifically, the set of people on the fedora-users and fedora-devel mailing lists are generally people who are already enamored of Fedora and/or have been working with it for a long time. This is traditionally a set of people who have established their own ways of working around (and sometimes mentally blocking) some of the more painful parts of the Fedora experience.
Put another way, limiting our source of input to the mailing list would be fundamentally equivalent to devising a project's budget estimate by only talking to the engineers. By not talking to marketing, quality-assurance, capital expenditures, facilities, etc., you'd come up with an inaccurate view.
Surveys might be a good idea, though someone would have to put them together, and someone would have to extract the information from them. Both of that isn`t easily done, either.
Right, and doing that properly (with an appropriate sample size as well as controls, such as users who have never used Fedora before) pretty much requires outsourcing to a professional research group, which costs $BIGNUM.
How do you currently find out what users want?
In my experience, we almost never find out what users want. We *often* hear after the fact what they don't like, but we rarely if ever hear recommendations ahead of time.
Furthermore, users are often REALLY BAD at knowing what they actually want. I've had people come to me and demand an option to disable SSL/TLS from some of my software. When they were pushed back on, it was ultimately discovered that they needed more transactions-per-second than they were getting and their simple experiments suggested that avoiding encryption would get it for them. We were able to do some profiling and bring the encrypted connections up to the performance they needed.
The moral here is that what they asked for and what they needed were not the same thing.
On 03/24/2014 04:15 PM, Liam Proven wrote:
On 24 March 2014 15:02, Ralf Corsepius rc040203@freenet.de wrote:
On 03/24/2014 03:12 PM, Liam Proven wrote:
On 24 March 2014 12:45, lee lee@yun.yagibdah.de wrote:
/usr belongs on it`s own partition. And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
I think that's a very 1980s, or early-1990s, way of looking at it.
C'mon, feeling something is oldfashioned is hardly an answer.
Well, actually, sometimes, yes, it is. Same as the decision to drop support for i386 from the kernel, or the fact that no installers default to ext2 any more.
These aren't old-fashioned, these are technically out-dated. Makes a huge difference!
Having been able to have /usr on a separate partition was a valuable feature, which now has gone lost. IMNSHO, ruined by naive, inexperienced kids (to use the same tone as you did), who were overwhelmed by the additional complexity supporting this feature had required.
I am not saying you're wrong, merely that I personally haven't seen a use or need for it since about 1989 and I found the reasoning for its collapse and merger to be sound.
The fact you haven't encountered it doesn't mean there are no use cases.
Just think about non-desktop HW (e.g. phones, tablets, routers, switches, NASes), which usually are equipped with different types of memory, being used for different purposes ("Linux as firmware").
Wrong. You are forgetting about systems booting from SD-Cards, USB-sticks and other forms of non-volatile memory.
Is Fedora a suitable OS for such hardware?
Definitely. Such setups are not uncommon on servers and are even sold by big brands. e.g. HP.
E.g. the HP ProLiant N36/40/54L - These are equipped with a built-in usb-2 socket, designated to take an USB-stick to boot the OS from.
Similar setups also aren't uncommon on HTPCs, NAS-boxes and similar boxes where "non-data partition"-filesystem performance is not of much importance.
Wrong. Most servers typically are headless, and if they have a graphic card-build-in, it's usually inaccessible or unused.
I am actually an IT professional - no, honestly, really I am - and every single rackmount server I've used in the last few years still has an SVGA port on it.
But is it used, is it really accessed? I guess no.
Also think about NASes or boxes being used as routers. No need for graphics on them.
Right, there is no strong necessity, nevertheless having these still would make sense.
There was an argument; it was decided not. I wasn't involved. I happen to agree, but I can't change it, so there's no point telling me! :¬)
I disagree.
Yeah, I guessed. :¬)
IMNSHO, UsrMove was a prominent epic fail in the long serious faulty decisions Fedora's leadership has committed.
We-eeeelll... I am not sure that I could overall disagree with the general thrust of your argument there. :¬)
No need to do so. RH has implemented facts which have rendered this discussion moot. IMO, some hidden cabal at RH had decided to pick the ancient (> 20 years old) idea to abandon separate partions for /usr and / and to sell it as "revolutionary novelty", instead of shooting it down such proposals as "Windows way of thinking", as it has been done for 20 years before :)
Ralf
Chris Murphy wrote:
Nowadays you may have SSDs which supposedly last longer when not written
much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
It's in the realm of 20+GB written per day every day, for the warranty period. If you're doing that, get an enterprise SSD. Or stick with HDDs.
A standard, non-enterprise, consumer SSD will last far longer[1] than anyone thinks. Please put the myths and conspiracy theories that /tmp or /var/tmp or anything on any SSD is "bad" to rest.
[1] http://techreport.com/review/26058/the-ssd-endurance-experiment-data-retenti...
On Mar 24, 2014, at 8:47 AM, Liam Proven lproven@gmail.com wrote:
On 23 March 2014 20:19, Chris Murphy lists@colorremedies.com wrote:
On Mar 23, 2014, at 12:24 PM, Liam Proven lproven@gmail.com wrote:
On 22 March 2014 18:25, Ed Greshko ed.greshko@greshko.com wrote:
Make sure you file a bugzilla.
How/why? It's not a bug.
How do you know? You haven't filed a bug report, you haven't presented any logs to anyone to find out if what you're experiencing is in fact intended behavior. You haven't even stated the nature of the installation failure.
I have to say that you are one of the most hostile, confrontational and aggressive people that I've come across in a support community since Richard Morrell of Smoothwall, about 12 years back, who alienated so many users that a bunch of them got together and forked it into IPCop... in large part so that they could have support communities without that man in them, telling them that they were stupid and wrong and should just go away if they didn't like it.
I'm the most hostile, confrontational, aggressive person you've come across in a support community in 12 years? Just like you haven't successfully installed Fedora to actual hardware in 10? Were you abducted by aliens for ~9 years by any chance?
Do you realise this?
Are you doing it deliberately?
Yes, I realize I'm more hostile than someone I've never heard of, and I'm doing it deliberately, sure.
It will not. Because this is the user to user forum. To cause a design change in the installer means posting to the Red Hat anaconda-devel@ list. To cause a design change specific to Fedora means posting to the Fedora devel@ list. Bugs, requests for enhancement, and even perceived clearly articulated design flaws should have bug reports filed. Pretty much always when an installer bug is filed, the installer logs should be attached (at least program.log and storage.log).
Well, OK. I am not a developer of or contributor to the product and since I can't even install it on my PC I am not very likely to become one, so I will just have to let it be for now, then.
You have put in more effort writing massively long emails over the past few days than it would take to find out the problem if you were so resistant to finding out the problem.
You're seriously acting like this is your first day at the rodeo. You are supposed to state what did happen,
What, like I did on the 5th of March in the threat entitled "Can't install ƒ20", when I said:
Fair enough. I just looked up that thread, and I see it rather quickly devolved into people bashing the installer rather than helping you solve the problem. So I'll argue they failed to be helpful other than to commiserate.
I will respond further in that thread rather than clutter this one up further.
including any error messages received
Tricky when the system hangs.
I don't know what that means exactly. You said the installer crashed with python errors. It's still possible to extract information from the system, short of a kernel panic which wouldn't be an installer problem.
and how this differs from your expectation of what should have happened.
What I expect? Well, in a nutshell, I expect to be able to say:
"Here, >this< is your root, >this< is /home, swap is >there<, put your bootloader in the root FS, now go."
Failing that, to be able to delete an existing root partition, make a new one, say ">that< is your root, go".
Is that so unreasonable?
Well it's really confusing because it's not a step by step in a way that I can reproduce. It's vague enough to match what I've done numerous times and has worked so there's some pre-existing configuration or step that you're doing that I'm not, and that's why it's not working for you and has for me. So you still need to provide more detail or logs or probably both.
The bootloader part isn't going to happen because since Fedora 18, embedding GRUB to a partition isn't supported. And it isn't supported because upstream GRUB folks don't recommend it, although this can be done to some degree post-install with the grub-install command and --force flag.
The installer always keeps logs in /tmp
As I said, it locked up.
Even if installer is hung, the bug reporter almost never is. If the bugreporter is, the kernel isn't and you can get to a shell to recover the logs. Did you try that? If you can't get to a shell, then this isn't only an installer problem.
- so go try a Fedora baremetal install and when it "won't install" and you're stuck, post the /tmp/anaconda.log, /tmp/program.log, and /tmp/storage.log.
I will have to look up what a "Fedora baremetal install" is. AFAICR I saw no such option in the boot/install image.
baremetal = to actual hardware, not a virtual machine
Chris Murphy
On 24 March 2014 16:11, Ralf Corsepius rc040203@freenet.de wrote:
On 03/24/2014 04:15 PM, Liam Proven wrote:
On 24 March 2014 15:02, Ralf Corsepius rc040203@freenet.de wrote:
Wrong. Most servers typically are headless, and if they have a graphic card-build-in, it's usually inaccessible or unused.
I am actually an IT professional - no, honestly, really I am - and every single rackmount server I've used in the last few years still has an SVGA port on it.
But is it used, is it really accessed? I guess no.
Server not responding, plug in monitor + keyboard and check it out. Often it will get rebooted anyway, but this is a useful tool. Even on a VM you can often connect to a display.
IMNSHO, UsrMove was a prominent epic fail in the long serious faulty decisions Fedora's leadership has committed.
We-eeeelll... I am not sure that I could overall disagree with the general thrust of your argument there. :¬)
No need to do so. RH has implemented facts which have rendered this discussion moot. IMO, some hidden cabal at RH had decided to pick the ancient (> 20 years old) idea to abandon separate partions for /usr and / and to sell it as "revolutionary novelty", instead of shooting it down such proposals as "Windows way of thinking", as it has been done for 20 years before :)
It turned out that usrmove was not a bad idea because it was 'windows thinking', but because it broke compatibility with other distributions and caused problems with applications that had been written relying on guarantees in the fs hierarchy. There are other people on this list who can fill in details on their battles with the results.
It doesn`t give you choices. It leaves you in the dark about that it is somehow possible to use an non-gui installer and to do a minimal install. It leaves you in the dark about what exactly happens when you do the partitioning and with trying to figure out how get the partitioning you want.
As I said before, the entire installer was rebuilt to promote a more laid back approach. The user can choose the order they wish to customize / experience the GUI installation instead of being forced down a specific path. There might be a couple mandatories, but for the most part it's all about choice.
A non-gui installation is not something that the majority of users will choose so it's not apparent, but if you want that method, here you go: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/ch-gui...
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
I definitely think the usability of the partitioning scheme in the current installer needs work, but as I stated earlier, I think some people are just griping about the change instead of there actually being an issue; although, it does sounds like you actually had an issue with it not detecting the correct size / free space of your drives. You might want to submit a defect, because in comparison, I have a multi-drive setup with LUKs encryption and I can have both drives wiped and start all over with encryption in less than a minute or if I want to retain one disk scheme but clobber the other it takes about 3 minutes.
A large majority of the information from the older installer is still there, it's just up to the user to seek it out.
I don`t know anything about "the older installer" or how to seek out information about it; I didn`t even know that there is an "older installer". The first Fedora installer I used was the one with F17.
F17 had the 'old' installer. The 'new' installer was introduced in F18.
In essence the installer went from a hardcore presentational format to a more laid back format - a shift every OS has consciously made in the last decade.
I don`t know what you mean. The Debian installer got more options and some more clarity which was an improvement. Otherwise it didn`t change, you just do country and keyboard setup --- which is missing in Fedoras installer, there was no way to tell it that I have a German keyboard --- network setup if you don`t use DHCP, partitioning, a bit of package selection if you want to, and then it installs.
Keyboard configuration is not missing... it's one of the main hub options: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/instal... http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/sn-key...
It`s easy and straightforward as it used to be for the last twenty years, and I never had trouble using it. Why suddenly make installing such a PITA like Fedoras installer does?
The computer industry knows that more and more people want installations to be less scary and faster. This trend has been seen in the evolution of the Windows installer, MacOS, most Linux distros, and even iOS or Android. There are going to be some Linux distros that don't embrace this, you mentioned Debian and I'm sure slackware and Arch won't either, but in the end it's all about attracting more users to the product. The choices are there, but us hardcore users just have to look more since we're the minority.
On Mar 24, 2014, at 9:57 AM, Liam Proven lproven@gmail.com wrote:
On 24 March 2014 15:51, Chris Murphy lists@colorremedies.com wrote:
Right, so in your view ten years of install failures to baremetal is an intentional feature?
And again with the challenging and hostility.
No, it's just a question. I'm trying to cut through the weeds to find out what I'm dealing with, with the ultimate goal of finding out why you're having difficulty installing Fedora.
I have not tried every single release; I tend to do distro roundups maybe every 3-5 years.
"To be brutally honest, even with commercial live-deployment Unix experience going back to 1989, I have /never/ got any release of Fedora to successfully install on real hardware. I find the installer that bad, and it's been getting worse, not better."
"As I have said previously, I have /never/ successfully installed Fedora on actual hardware since v1.0 shipped in, what was it, 2003?"
Based on what you've said, in particular the use of "any release", a reasonable person can conclude you've tried them all. In that context, don't you think that 20 failures to install seems pretty nutty?
So now that the context is clearly rather different, it make slightly more sense.
The particular problem with F20 is one I've never seen before, in Fedora or in any other OS ever.
I understand that you're not familiar with some of Fedora's rough edges, the fact the installer code is very new as a result of a ground up rewrite just a year ago, and by signing up for Fedora you are indirectly signing up to be a tester to some degree.
How did you find out about the list and how did you sign up for it? Doing several Google searches for Fedora user lists, the top most result is this page which makes it abundantly clear there are many Fedora lists.
Uhuh. And from this I am meant to *just know* by my powerful psychic powers who reads it or not, am I?
Am I wearing clown shoes?
No, by inference.
Oh so that's a reason to not even try?
Actually, yeah.
And that is why you fail to install Fedora. At this point it's every bit as much user failure as it is installer failure in my opinion.
As I have said, you can document it. I've told you where to find the logs, and more than one way to make them available. Your lack of curiosity and active resistance to finding out the cause of these install failures is suspicious.
No, I think *you're* suspicious, and that is a rather different thing.
I am suspicious why someone would devote this many emails, ostensibly asking for help, while being so magnificently circumspect when it comes to actually being helped.
When it comes to providing the most basic information to help solve this problem, you resist it. What do you want? Do you want to get Fedora to install on real hardware or not? Or do you just want to gripe about the installer for another 25 emails?
Chris Murphy
On Mar 24, 2014, at 10:14 AM, Michael Cronenworth mike@cchtml.com wrote:
Chris Murphy wrote:
Nowadays you may have SSDs which supposedly last longer when not written
much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
It's in the realm of 20+GB written per day every day, for the warranty period. If you're doing that, get an enterprise SSD. Or stick with HDDs.
A standard, non-enterprise, consumer SSD will last far longer[1] than anyone thinks. Please put the myths and conspiracy theories that /tmp or /var/tmp or anything on any SSD is "bad" to rest.
Yeah I agree there's too much nervousness about SSD wear issues. And actually the better/best consumer SSDs can tolerate maybe twice that amount so before going enterprise SSD, even look at a higher quality more expensive consumer SSD.
Chris Murphy
On 03/24/2014 10:38 AM, Tom Horsley issued this missive:
On Mon, 24 Mar 2014 11:27:14 -0600 Chris Murphy wrote:
Yeah I agree there's too much nervousness about SSD wear issues.
And, of course, hard disks just never fail in comparison, right? :-).
It's not so much an issue that hard disks don't fail, but they usually start giving some indication they're having issues (retries, etc.) so you can plan for it (get a new drive, make sure you have a backup so you can replace the drive and recover).
SSDs just...die. No warning, nothing. Suddenly the machine won't boot and you find the SSD is just a lump of useless silicon. I've had to replace at least four SSDs in Macbooks in the last year. Fortunately we had backups for three of them, but the fourth guy was adamant that he didn't need backups. Needless to say he was rather angry when it crapped out on him. "I hate to say we warned you, but......" ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks@alldigital.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - Sarchasm: The gulf between the author of sarcastic wit and the - - reader...who doesn't get it. - ----------------------------------------------------------------------
On Mon, 2014-03-24 at 14:36 +0000, Liam Proven wrote:
On 24 March 2014 01:33, Patrick O'Callaghan pocallaghan@gmail.com wrote:
On Sun, 2014-03-23 at 18:24 +0000, Liam Proven wrote:
On 22 March 2014 18:25, Ed Greshko ed.greshko@greshko.com wrote:
Make sure you file a bugzilla.
How/why? It's not a bug.
If it doesn't work as it's supposed to, it's a bug.
An intentional design feature that causes a problem in an unexpected or unconsidered use-case is not what I'd call a bug, but hey, have it your own way. :-)
I think you're reaching if you *don't* think that's a bug. It's not a question of me having my own way, but of following the usage on this list.
Talking about it on this list will not result in any action.
Won't it? Why not? Are there multiple non-overlapping user communities for Fedora, then? A problem is a problem; where it is reported shouldn't matter, should it?
Seriously? Yes, there are multiple partially-overlapping communities.
Yes seriously. I'm new here. I have no idea who does or does not read this.
Fine, then take on board the fact that the Users list is for discussing the current and previous release of Fedora, and naturally that includes the problems people have with it, but if the outcome of a discussion is to identify a bug it's not enough to just leave it at that. To not have it forgotten about then you need to use Bugzilla.
Many of the developers don't read this list. Many of the people on this list don't read the Test list. If you want to document a bug then use Bugzilla.
As I have said, I don't think that I /can/ document it well enough to be of any use.
"Document" simply means "describe with as much detail as possible".
poc
Once upon a time, Tom Horsley horsley1953@gmail.com said:
On Mon, 24 Mar 2014 11:27:14 -0600 Chris Murphy wrote:
Yeah I agree there's too much nervousness about SSD wear issues.
And, of course, hard disks just never fail in comparison, right? :-).
Of course they do. The difference is that traditional hard drives don't generally have a lifetime measured by number of writes. Hard drives have decades of institutional experience, while SSDs (and their failure modes) are new and different. And, new (and no experience) is always "scary".
With traditional hard drives, failure modes are mitigated by RAID (to keep the system running) and backups (to protect the data from drive and other types of failure). With any storage, you should still do backups, but for a while RAID setups didn't pass through TRIM commands and so were not recommended for SSDs. That took away a "warm fuzzy feeling" for storage administrators, and the fact that some (many?) of the early consumer SSDs were not terribly reliable didn't help.
Now, I have SSDs in all my home computers (desktop, server, notebook, and HTPC). I still have "spinning rust" hard drives in my network disk server, because I don't want to pay for 3TB of SSD (and hard drives are plenty fast for that type of storage).
On Mon, 24 Mar 2014 11:30:23 -0700 Rick Stevens wrote:
It's not so much an issue that hard disks don't fail, but they usually start giving some indication they're having issues
Not any disk I ever had fail. All of 'em worked perfectly right up to the instant where they wouldn't boot one day. Nothing in the logs, no smart warnings, nuthin.
In fact, I've always wondered what good all the smart reporting is for since no disk I've owned has ever had any problems reported right up to the point where it wouldn't talk enough for smart to access it :-).
On Mar 24, 2014, at 12:36 PM, Chris Adams linux@cmadams.net wrote:
Once upon a time, Tom Horsley horsley1953@gmail.com said:
On Mon, 24 Mar 2014 11:27:14 -0600 Chris Murphy wrote:
Yeah I agree there's too much nervousness about SSD wear issues.
And, of course, hard disks just never fail in comparison, right? :-).
Of course they do. The difference is that traditional hard drives don't generally have a lifetime measured by number of writes. Hard drives have decades of institutional experience, while SSDs (and their failure modes) are new and different. And, new (and no experience) is always "scary".
With traditional hard drives, failure modes are mitigated by RAID (to keep the system running) and backups (to protect the data from drive and other types of failure). With any storage, you should still do backups, but for a while RAID setups didn't pass through TRIM commands and so were not recommended for SSDs.
In the realm of "we worry too much about SSD wear" is trim, a.k.a. discard mount option or the use of fstrim command.
It's actually a lot more important for things like LVM thinp and RAID to be "trimmable" than the underlying physical device. Use of trim on the physical device is immediately problematic if the device only returns non-deterministic data after trim. It needs to support either DRAT or DZAT, or maybe both (I forget). Otherwise, whether raid1, 5, or 6, the return of non-deterministic garbage results in mismatches that are indistinguishable from massive corruption when doing a regularly scheduled scrub. So yeah, not good.
smartctl --identify=wb /dev/disk0 | grep -i trim
69 14 0 Deterministic data after trim supported 69 5 0 Trimmed LBA range(s) returning zeroed data supported
That's on a Samsung 830. I'm not positive, but that looks like neither are supported on this SSD. So that likely means:
1. RAID 0 + trim = OK. 2. RAID 1 + no trim = OK, scrub reliable. 3. RAID 1 + trim = Is redundant but scrub unreliable, will return many mismatches.
An exception is Btrfs raid1 + trim, because it only scrubs allocated data. It doesn't scrub by block.
Chris Murphy
On Mar 24, 2014, at 1:04 PM, Tom Horsley horsley1953@gmail.com wrote:
On Mon, 24 Mar 2014 11:30:23 -0700 Rick Stevens wrote:
It's not so much an issue that hard disks don't fail, but they usually start giving some indication they're having issues
Not any disk I ever had fail. All of 'em worked perfectly right up to the instant where they wouldn't boot one day. Nothing in the logs, no smart warnings, nuthin.
I think it's something like 60% for the prediction of SMART failures and often you have maybe minutes to do something about it. Anyway, SMART definitely permits a lot of drive failures without prior warning.
That percentage goes up a lot if you look at all of the attributes, and choose to replace based on attributes trending with higher raw values (trending lower normalized value) that are associated with prefail. But I think it was the Google study that found even looking at all attributes still wasn't enough of the right kind of data to do really useful failure prediction.
In fact, I've always wondered what good all the smart reporting is for since no disk I've owned has ever had any problems reported right up to the point where it wouldn't talk enough for smart to access it :-).
The SMART overall-health self-assessment test result is next to useless. It's basically a pass/fail. A better picture emerges when looking at the individual attributes. Still it's not fool proof because all you're doing is somewhat subjectively saying "OK if bad sectors start to trend upward at a rate greater than, say, 1 per week, I'm replacing the drive." I don't know that anyone has a clear line in the sand on the attributes though, there's a huge gray area.
Chris Murphy
On 03/24/2014 05:24 PM, Ian Malone wrote:
On 24 March 2014 16:11, Ralf Corsepius rc040203@freenet.de wrote:
On 03/24/2014 04:15 PM, Liam Proven wrote:
On 24 March 2014 15:02, Ralf Corsepius rc040203@freenet.de wrote:
Wrong. Most servers typically are headless, and if they have a graphic card-build-in, it's usually inaccessible or unused.
I am actually an IT professional - no, honestly, really I am - and every single rackmount server I've used in the last few years still has an SVGA port on it.
But is it used, is it really accessed? I guess no.
Server not responding, plug in monitor + keyboard and check it out. Often it will get rebooted anyway, but this is a useful tool.
Well, I did not say, "There are no servers w/ GPU". All I said, is the assumption "all servers were having a usable/accessible graphics is invalid".
Sometimes the HW doesn't have a GPU, sometimes MoBo is physically inaccessible/hardly inaccessible, sometimes the GPU is switched off/disabled, sometimes nobody cared to configure the GPU.
On the other hand, I've also seen cases, where switching on the GPU and running runlevel 5 on servers helped to reduce power consumption ;)
IMNSHO, UsrMove was a prominent epic fail in the long serious faulty decisions Fedora's leadership has committed.
We-eeeelll... I am not sure that I could overall disagree with the general thrust of your argument there. :¬)
No need to do so. RH has implemented facts which have rendered this discussion moot. IMO, some hidden cabal at RH had decided to pick the ancient (> 20 years old) idea to abandon separate partions for /usr and / and to sell it as "revolutionary novelty", instead of shooting it down such proposals as "Windows way of thinking", as it has been done for 20 years before :)
It turned out that usrmove was not a bad idea because it was 'windows thinking', but because it broke compatibility with other distributions and caused problems with applications that had been written relying on guarantees in the fs hierarchy. There are other people on this list who can fill in details on their battles with the results.
;) I have been (and occasionally still am) involved with struggling with the negative impacts of /UsrMov - These battles are not over, yet.
My primary critism on /UsrMov is the way Red Hat had pushed it. I regret having to say so, but to me, this was a rude act, accompanied by a disinformation propaganda campaign.
Ralf
On Mon, Mar 24, 2014 at 7:12 AM, Liam Proven lproven@gmail.com wrote: ...
There is no need for separating out admin binaries, user binaries, local binaries, graphical binaries etc. any more, and hasn't been for about 2 decades.
What about DRBL (Diskless Remote Boot Linux) and LTSP (Linux Terminal Server Project)? If /usr is nfs mounted then booting becomes an issue.
It's important to remember that there are *many* different but valid use cases. Similarly, I think Systemd is another big disruption for these projects.
Chris Murphy lists@colorremedies.com writes:
On Mar 24, 2014, at 6:45 AM, lee lee@yun.yagibdah.de wrote:
/usr belongs on it`s own partition.
As if no one has ever said that before, and as if it convinced even one thinking person to change their mind.
Thinking persons do not need to change their minds about it because they realise that being able to have /usr on it`s own partition is a good thing.
Fedora has never defaulted to separate /usr partition. It's been two years since this was decided. That you're still experiencing cognitive dissonance over this ancient long ago resolve topic is your problem, not anyone else's.
I am not experiencing cognitive dissonance, and it`s not my fault when Fedora made a retarded decision before I even started using it. This discussion is about Fedora.next, and it might be possible to fix the problem in some future version of Fedora.
And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
bin lib lib64 are symlinks to their locations in /usr.
And that is supposed to be a good thing?
Bill Oliver vendor@billoblog.com writes:
Much of what we consider very important today was considered stupid when it first came out.
Like?
Well, the automobile, aircraft, and personal computer immediately come to mind.
Those are good examples. Yet nobody would take away your horse and cart just like that and replace it with an automobile.
Chris Murphy lists@colorremedies.com writes:
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
Please post the bugzilla URL.
I didn`t make one, I was busy trying to install. You need to get used to that not every problem is or can be made into a bug report, especially not one that would be in any way useful.
I already said what I would have put into such a bug report in another post. It won`t be useful.
Most of the size reporting problems like this are non-contiguous sections of free space being added up and reported as Available space; but the request is for a partition size greater than the largest contiguously available space.
Maybe it begins with the installer messing together all the disks in some weird way rather than to treat them separately and just let you partition them the way you want to. IIRC, there wasn`t even a way to tell it which partition to put where. You could either pick only disk (of the two there were) and install on only that, with the the other one completely unusable. Or you could pick both of them, in which case there was no way to tell what would happen because they were messed together somehow. You didn`t have a choice about the swap partition, either. I would just force you to use one that was already existing, with no way to tell it not to use it or to delete it. It just sucked.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
It isn't going to get better complaining about it on this list. Do you have bugzilla IDs, and if so post them. If not, then how do you expect the behavior to get any better? Magic?
The makers of the installer can always look into this list and see what ppl say about the installer and learn from that. Bug reports are not suited for this, and complaining that ppl don`t make enough of them doesn`t get you anywhere.
Or, since you keep insisting on bug reports, why don`t you go ahead and put together a list of URLs to the list archive pointing to posts about the installer to compile a combined bug report?
How much attention and fixing do you think that would get? It`ll probably be closed with WONTFIX, one reason being that what is said here doesn`t refers much more to installers from F17 to F19.
Perhaps the installer of F20 has be re-designed from scratch, at least when it comes to partitioning, and now works fine.
you just do country and keyboard setup --- which is missing in Fedoras installer, there was no way to tell it that I have a German keyboard ---
Fedora 18, 19 and 20 have a keyboard spoke in the installer which is how you tell it you want to use a German keyboard layout.
Whatever they might have, I tried several times (because I had to start over many times because it refused to do the partitioning) to tell the installer of F19 that I have a German keyboard, and there was no way. I don`t even know what you mean by "spoke". You boot the life system, search for "install" to find the installer, then you get an icon and start the installer, and pretty soon you get stuck with trying to do the partitioning.
It`s easy and straightforward as it used to be for the last twenty years, and I never had trouble using it. Why suddenly make installing such a PITA like Fedoras installer does?
Please don't ask silly questions that propose the intended design goal was to piss users off, it's irritating.
That seems to be an assumption you have.
If Debian has changed their installer to something the like what Fedora has, I really don`t see why they suddenly would waste their effort on creating a new installer that makes installing a PITA rather than keeping and improving a perfectly good installer. So tell me: Why would they?
I wouldn`t mind if they made a new installer as good or even better as the existing one. They`d only need to leave the existing one around, possibly as an easy to find option, or as the default, until the new is ready to replace it.
Liam Proven lproven@gmail.com writes:
On 24 March 2014 12:45, lee lee@yun.yagibdah.de wrote:
/usr belongs on it`s own partition. And last time I looked, it would not be compliant with the FHS not to have what is needed in /bin and /sbin but to use symlinks instead.
I think that's a very 1980s, or early-1990s, way of looking at it.
Since the normal way to boot a PC now is a complete functioning OS on a single removable-media volume - be that an optical disk or USB flash media
I guess you mean when booting an installer?
- most of the rationale for splitting up the bits of the /usr
tree have long ceased to apply.
Which ones in particular are you referring to?
The smallest hard disks available today (~500GB) are roughly 2 orders of magnitude bigger than is needed for a full Linux desktop install (~5GB). It is not possible to buy a new computer without a graphical display.
And?
Besides, you can still buy disks with less capacity.
There is no need for separating out admin binaries, user binaries, local binaries, graphical binaries etc. any more, and hasn't been for about 2 decades.
I have always found it very useful, and it still is. Are you even assuming that people have only single disk or single volume?
I think it's a brilliant, if brave, idea of Fedora to get rid of a historical distinction that is now pointless, but it's planned and discussed and decided, as far as I know:
I don`t exactly know what they are doing, but I find it would be a very stupid idea to give up a useful and reasonable distinction between software for different uses and to loose the order the FHS brings about. Sure you can do it as an option if you like and put everything on a single partition; nobody keeps you from doing that. Don`t force it upon anyone who doesn`t want it, though.
Michael Cronenworth mike@cchtml.com writes:
Chris Murphy wrote:
Nowadays you may have SSDs which supposedly last longer when not written
much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
It's in the realm of 20+GB written per day every day, for the warranty period. If you're doing that, get an enterprise SSD. Or stick with HDDs.
A standard, non-enterprise, consumer SSD will last far longer[1] than anyone thinks. Please put the myths and conspiracy theories that /tmp or /var/tmp or anything on any SSD is "bad" to rest.
That`s why I said "supposedly". I don`t have an SSD and haven`t seen any conclusive data yet that would actually show which type of disk lasts longer.
Liam Proven lproven@gmail.com writes:
Having been able to have /usr on a separate partition was a valuable feature, which now has gone lost. IMNSHO, ruined by naive, inexperienced kids (to use the same tone as you did), who were overwhelmed by the additional complexity supporting this feature had required.
I am not saying you're wrong, merely that I personally haven't seen a use or need for it since about 1989 and I found the reasoning for its collapse and merger to be sound.
And that you can`t see a use or need for something means that it`s obsolete?
"Powell, Michael" Michael_Powell@mentor.com writes:
It doesn`t give you choices. It leaves you in the dark about that it is somehow possible to use an non-gui installer and to do a minimal install. It leaves you in the dark about what exactly happens when you do the partitioning and with trying to figure out how get the partitioning you want.
As I said before, the entire installer was rebuilt to promote a more laid back approach. The user can choose the order they wish to customize / experience the GUI installation instead of being forced down a specific path. There might be a couple mandatories, but for the most part it's all about choice.
Which installer are you referring to? As to Fedora, I have only used the ones of F17 and of F19, and none of them gave me any choices. You can start the installer and it doesn`t even let you do the partitioning you want, which is the only choice you are getting.
I`m not an expert with Fedoras installers in any way. This is simply my "user experience". Maybe the "user experience" the installer provides should be different.
A non-gui installation is not something that the majority of users will choose so it's not apparent, but if you want that method, here you go: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/ch-gui...
Says who? And why not give the users that choice instead of hiding it?
Partitioning took me about three hours with the installer of F19, with a very
[...]
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
I definitely think the usability of the partitioning scheme in the current installer needs work, but as I stated earlier, I think some people are just griping about the change instead of there actually being an issue;
Well, I don`t know what changed.
although, it does sounds like you actually had an issue with it not detecting the correct size / free space of your drives.
It did detect the disks and showed them, but it didn`t let me do what I wanted.
You might want to submit a defect, because in comparison, I have a multi-drive setup with LUKs encryption and I can have both drives wiped and start all over with encryption in less than a minute or if I want to retain one disk scheme but clobber the other it takes about 3 minutes.
Sooner or later I`ll probably try installing on a software RAID-1 with encryption, with nothing else on the disks. Can I make a video of that by recording what`s on screen, after booting the life system? It would be possible to store the video over network.
A large majority of the information from the older installer is still there, it's just up to the user to seek it out.
I don`t know anything about "the older installer" or how to seek out information about it; I didn`t even know that there is an "older installer". The first Fedora installer I used was the one with F17.
F17 had the 'old' installer. The 'new' installer was introduced in F18.
I like the old one better. Except for not letting me have a /usr partition, it worked.
In essence the installer went from a hardcore presentational format to a more laid back format - a shift every OS has consciously made in the last decade.
I don`t know what you mean. The Debian installer got more options and some more clarity which was an improvement. Otherwise it didn`t change, you just do country and keyboard setup --- which is missing in Fedoras installer, there was no way to tell it that I have a German keyboard --- network setup if you don`t use DHCP, partitioning, a bit of package selection if you want to, and then it installs.
Keyboard configuration is not missing... it's one of the main hub options: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/instal...
That one looks rather unfamiliar. Was that in the F17 or F19 installers?
It`s easy and straightforward as it used to be for the last twenty years, and I never had trouble using it. Why suddenly make installing such a PITA like Fedoras installer does?
The computer industry knows that more and more people want installations to be less scary and faster. This trend has been seen in the evolution of the Windows installer, MacOS, most Linux distros, and even iOS or Android. There are going to be some Linux distros that don't embrace this, you mentioned Debian and I'm sure slackware and Arch won't either, but in the end it's all about attracting more users to the product. The choices are there, but us hardcore users just have to look more since we're the minority.
Faster, ok, that mainly depends on what the bandwidth of your internet connection is and what the server side delivers. The only two scary things are the possibility that the network device doesn`t work (like Debian missing modules for some) and, far worse, potentially loosing your data.
The Fedora installers I used were fine with network devices and extremely scary with the not loosing your data part.
Having more choices doesn`t make installing any slower or any more scary. It makes it easier. Not having choices makes installing much slower and more scary.
Chris Murphy lists@colorremedies.com writes:
On Mar 23, 2014, at 3:56 PM, lee lee@yun.yagibdah.de wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
Old news.
https://fedoraproject.org/wiki/Features/UsrMove#I_have_.2Fusr_as_a_separate_...
and
http://www.freedesktop.org/wiki/Software/systemd/separate-usr-is-broken/
So Fedora failed miserably because these articles point out that /usr can be mounted read-only and assume that /usr can be on it`s own partition, maybe shared over network.
And following the arguments for moving everything to /usr, you would have to say that /usr should be in /boot. You can then use symlinks from there.
RAID isn`t exotic, either.
It kinda is. The definition of exotic is "not ordinarily encountered." And that's even if you look at just the Linux universe, because overwhelmingly most users don't use it. If you look at the rest of the computing world it's either not an install time option or not possible.
It`s a a recipe for failure not to use raid, and which Linux distribution doesn`t support it? It`s "ordinarily encountered" not only with Linux distributions. Perhaps an overwhelming majority of users use it, you don`t know that. You could as well say that being able to define styles for paragraphs with WYSIWYG word processors is an exotic feature because the great majority of users don`t use it.
I always use separate partitions. It has lots of advantages.
It also has many disadvantages. Hence LVM thinp and btrfs subvolumes as alternatives.
Nobody says that you have to use either --- except the F17 installer not only telling you but not even allowing you to have /usr on its own partition like it should be.
I might split off /var on a server but I'd need a remarkably persuasive use case, and on servers, I use extra-stable distros without GUIs, not something like Fedora.
/var can get full, and it`s written to, same goes for /tmp.
Use quotas.
Why the hassle? And quotas don`t magically adjust the performance of the disks.
Nowadays you may have SSDs which supposedly last longer when not written much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
It's in the realm of 20+GB written per day every day, for the warranty period. If you're doing that, get an enterprise SSD. Or stick with HDDs.
How long the warranty lasts is pretty irrelevant.
Why wouldn`t you use different partitions? I can see it (and have done it) for when the available disk capacity is extremely limited, but otherwise it doesn`t make any sense and has nothing but disadvantages.
It's cute when people project their world view in such narrow terms. Looking over at Windows and OS X, for 20+ years they've had at most two partitions, one for boot and one for everything else. OS X only just went from one partition for everything to two just a few releases ago in order to support full disk encryption (around a decade after Linux had it, but then it's also on-the-fly COW online convertible bidirectionally). So there isn't an inherent good for partitioning. It's useful for certain use cases. It's a negative for others.
Windoze is a total mess. Try to keep the applications you have installed separate from the data and the system and you`ll see. I don`t know what OS X does. Sure you can mess everything together without using partitions. That doesn`t mean it`s a good idea.
But this just illustrates the breadth of scenarios a successful installer must cope with!
It`s merely a reasonable standard thing to use separate partitions and a requirement to use RAID, and encrypted partitions for laptops, not something in any way unusual.
It is in fact unusual. What you're doing is proposing that what works for you, as a default for everyone. And your arguments for changing the paradigm are completely uncompelling.
There`s nothing unusual about it. You may have done things differently, and whatever reasons you might have for that are not convincing. That doesn`t mean that everyone should be forced to do it the same way as you do.
"a requirement to use RAID" is particularly irritating because you're saying everyone without boot from SAN capability, or a laptop, should be required to have two like sized drives to do an install. Otherwise it wouldn't be "required". So what you're writing doesn't even make sense.
I`m not saying that. Disks fail, and especially when you have your data on them, it`s a hassle when it happens and you might lose data. Backups only help so much. If you don`t care about your data and/or don`t need reliability, don`t use RAID, you`d make things unnecessarily complicated and be wasting your money. When you want to keep your data and don`t want the hassle with failing disks, RAID is a requirement. Or do you know something better?
The SAN device you`re booting from doesn`t store data in thin air, does it?
The Fedora installer will use existing layouts: partitions, PV/VGs, Btrfs, and md RAID. Click on the existing logical device and then specify a mount point for it.
Like I said, it finally worked after I created the partitions otherwise. It still wasn`t easy to tell it what to use for what.
Why not give them a choice, like either cfdisk or parted, then tell the installer what to do with each partition and let them switch between these until they are done
Use kickstart?
What is kickstart?
I think letting users switch between two tools means both tools are failures, or the user is neurotic.
Each user has preferences. Nobody says that when you can choose between cfdisk and parted that you are forced to use both. You can simply use the one you prefer.
I don`t remember what I used to create the partitions when Fedoras installer didn`t work, it was either cfdisk or parted and it doesn`t matter. I did have to switch between tools, i. e. the installer and a partitioning tool. By your logic, both tools were failures, or a user for whom the installer doesn`t work is neurotic.
Your logic is flawed. The partitioning tool did it`s job fine. The installer failed to do that part of its job, and I`m not neurotic.
Anyway, it is not a failure to give users choices. It`s merely a week point of Fedora that it doesn`t give choices and that it hides those that there are from the users.
Pick a tool for a task, move on to the next tool. Switching back and forth isn't going to be stable. You're talking about a tiny percent of users doing what you suggest, which means an infinitesimal number are testing it.
Seldom used or seldom tested things have no place in a GUI installer.
Try the Debian installer.
Liam Proven lproven@gmail.com writes:
On 23 March 2014 21:56, lee lee@yun.yagibdah.de wrote:
There`s nothing weird or exotic about it. I`ve always had /usr on its own partition until the F17 installer refused that, which it shouldn`t have.
As I have commented elsewhere, I think this is a 1980s style of thinking. Things have changed. Move on. Sorry, but they have; see the links elsewhere.
It doesn`t matter what you think. I have good reasons to do the partitioning I do.
Besides, I would find it unreasonable to say that something which has been done in a particular way since and for a long time is suddenly something weird and exotic.
RAID isn`t exotic, either. Disks do fail, the only question is when, and I neither want to lose data, nor the hassle.
Sure, I use it on all my servers.
Installing on a laptop requires encrypted partitions. They can be stolen too easily.
I have never ever used this and never expect or plan to. I suggest that your blanket statement is too sweeping.
You don`t need to plan or to expect to ever use something. Perhaps you have a better alternative to prevent data stored on laptops from getting into the wrong hands for instances when the laptops get stolen. Perhaps you always do messy installs without decent partitioning. Perhaps you don`t --- it doesn`t matter because you are not the only person for whom Fedoras installer would need to work.
I never separate out /tmp or /var or /usr/local - I only ever use / and /home basically.
I always use separate partitions. It has lots of advantages.
As I said, I use / and /home and advise against combining them.
And I advise against it. Do it when you like it and let others do what they like. I could very well claim that not using separate partitions is weird and exotic, just as you claim that using them is.
Personally I think that's enough. I am not disputing your reasons, but AIUI, Fedora is trying out a move to flatten and simplify the way-too-complex directory hierarchy. It's happened. The decision is made. Deal with it, move on.
There is nothing too complex about it. Should I find that I still can`t have /usr on its own partition next time I install Fedora, I`ll just install something else instead.
/var can get full, and it`s written to, same goes for /tmp.
As I said elsewhere: when the smallest new HD you can buy is half a terabyte (and even SSDs start at half that) this really isn't a big issue any more.
So when a disk has a capacity of at least 500MB, it never can get full. Please prove this theory.
How do you mount /usr read-only?
I don't. Never have in 26y of Unix systems support. For rescue, now, I boot off a LiveDVD or LiveUSB.
I do. It`s a very simple and useful precaution and weird and exotic not to do it.
Especially on a server, it`s a good idea to mount everything read-only that you can. When you have several disks, you can do your partitioning in such a way that you get better performance.
[...] So he came in and I set up a test and showed him that, to 2 decimal places in a percentage-based benchmark score, i.e. well below measurement error, there was absolutely *no* difference between the speeds of different areas of the disk.
I never actually tried to test it, so this may be the case or not.
Putting things at different places of a disk it`s not the only way to make use of partitions to get better performance.
Now, it is not real. It is not there any more. Disks are a thousand times bigger and faster now. This stuff does not matter any more and hasn't since before Linux 2.0 was released.
You are omitting that the amounts of data also have increased.
Especially when you have a server, you may need a (pretty much) granted capacity on /var or /tmp to make sure it will continue to operate --- without separate partitions, your users may fill up the disks ...
Users shouldn't be able to write stuff to / at all! Only to /home or below, or dedicated data partitions. Where /usr or /var is should not matter to them.
I didn`t say that users should be able to write to /. When you don`t have a separate partition for /home, they don`t need to write to / to fill it.
Log files and cached files go into /var and may fill it.
Nowadays you may have SSDs which supposedly last longer when not written much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
Machines come with dozens of gigs of RAM now. I'm not sure there's much argument for swap at all, and personally, I use tmpfs for better performance and a self-cleaning /tmp tree.
It`s nice for you when you have one of those. They can still run out of memory, and swap can be used to slow them down so much that you may be able to do something before arbitrary processes are killed and the whole system may go down.
Why wouldn`t you use different partitions? I can see it (and have done it) for when the available disk capacity is extremely limited, but otherwise it doesn`t make any sense and has nothing but disadvantages.
Exactly. One splits stuff up when space is an issue. When it isn't, one doesn't need to.
You got that wrong, it`s the other way round.
It`s merely a reasonable standard thing to use separate partitions and a requirement to use RAID, and encrypted partitions for laptops, not something in any way unusual. Of course I expect an installer to handle that as well as using a single, unencrypted partition on a single disk.
Actually, I agree, but there must be /some/ limits to the granularity!
You can have reasonable ones, but technically they are probably not required.
And it`s not too difficult. The installer doesn`t need to do the partitioning, the user does it. The installer only needs to give the user a good tool to do the partitioning the user wants and let them use it. Good tools to do partitioning are already available, and the installer doesn`t need to re-invent the wheel in that.
I mostly agree, but bear in mind that the installer must cope with both experts and novices. That's a tough call.
That`s what the "just install" option I mentioned is for.
I'd say it's /too/ simplified at the moment, though, and I think you might agree...?
Getting the partitioning you want with Fedoras installer is anything but simple. I can only speak for the installers of F17 and F19, though. With the F19 one, it was impossible.
Perhaps it even shouldn`t. Why force the user to learn how to use yet another partitioning tool they even rarely use unless they install Fedora all the time? Why not give them a choice, like either cfdisk or parted, then tell the installer what to do with each partition and let them switch between these until they are done --- or let the installer do whatever partitioning it wants, which means that all existing data on the disks will be deleted.
Mostly, I would agree, actually.
It may need a third option: Use available free space only and don`t delete anything.
On Mon, Mar 24, 2014 at 08:19:52PM +0100, lee wrote:
/usr belongs on it`s own partition.
As if no one has ever said that before, and as if it convinced even one thinking person to change their mind.
Thinking persons do not need to change their minds about it because they realise that being able to have /usr on it`s own partition is a good thing.
It's important to realize that you *can* have a separate /usr -- it just really needs to be available at boot time. That means you can have separate mount options, filesystems, partition constraints, or whatever. It just doesn't work anymore to have it on a network share or (if anyone ever did this!) removable media added after initial boot.
I used the network share case in the mid 1990s, when we were trying to cram Irix 6 onto 800MB workstation drives. These days, that's not really an issue. (And, hey, you can fit minimal Fedora in that space!) It might be neat for some special cases, but I hope we can all agree that it *is* a special case (and that Fedora isn't necessarily the right thing to cover all special cases).
Tim ignored_mailbox@yahoo.com.au writes:
I reckon it's the case for most OSs that /most/ users don't really care much about what they're using, nor how it works. The large number of clueless people using computers would seem to be evidence of that.
The number of clueless people is also large without computers. Computers might only make that number larger.
It's a lesser number of people that have concerns about their OS. Maybe that's why developers are less concerned about public opinions.
Or maybe people don`t bother to have or bring about concerns because the developers don`t care.
Stephen Gallagher sgallagh@redhat.com writes:
On 03/24/2014 09:22 AM, lee wrote:
The ones making packages probably have more influence. Is it supposed to be like that?
Frankly, yes. Feedback on a list is fine, but anyone can say "Hey, I wish it was more like this:", but ultimately it will be up to someone to actually implement that change. The ones who go and do the work are the ones who have the final say on what happens.
Sure, and the ones who do the work can`t very well ask the users for every little detail of an implementation. Still that someone needs to do the work doesn`t mean that the ones doing it /should/ have more influence by default.
Very little change or improvement ever happens because a lot of people talked about doing something for years. Things change because someone actually goes and makes it happen. That's the culture we try to encourage in Fedora.
Then it is irrelevant what the users think unless they care to and manage to find some way to make it happen --- and once they do that, they aren`t users anymore. It means that users are not involved in making a distribution and not concerned with things like leading the advancement of FOSS.
This makes the whole discussion about Fedora.next off-topic here, doesn`t it?
That's why we have Gnome, KDE, LXDE, and MATE-Compiz desktop spins -- and, pointedly, not a fvwm one. If you really think that this is the best course for Fedora, I encourage you to step up and create one. (See https://fedoraproject.org/wiki/Spins_Process.)
I don`t understand why every possible choice should require making a distribution on it`s own.
You're confusing a spin with a distribution (and a product with a distribution).
Perhaps --- I know I can download different life systems to install Fedora which come with different "desktop environments". That presents itself as a bunch of different distributions which are all Fedora, with one of them apparently being the "original" and the only apparent difference being that they install different "desktop environments" by default. Since I`m not using any of them, it doesn`t matter which distributions I pick, so I picked what seems to be the "original", vaguely guessing that it might be the most complete and perhaps best thought out one, and go from there.
Whether one or some of these distributions is or are called "spin" or "product" is irrelevant for me. It only makes me wonder why I can`t download an installer, preferably a life system, and install from there, choosing what I want to have when I`m about to install. Will I miss something and/or not be able to have it when I download the installer for the "original", or for one of the others?
That`s merely "user experience". Why make it confusing for the user?
Both products and spins are curated sets of packages from a single distribution (Fedora). Each one has its own reason for existing (in the case of the desktop spins, it's basically to show off a particular piece of technology).
Why would that require it`s own distribution? Can`t I just choose to install that particular piece of technology when I want to try it out?
For the Products, we're working to establish specific *solutions*. Recognizing that most people install an operating system so that, well, they can operate their system, we're trying to build solutions for three common use-cases so that newcomers to the Fedora Project don't feel like they need to make a thousand individual package choices to get their system running.
Why would that require different distributions? Just have something like what Debian calls "tasks", i. e. particular pre-defined package selections users can choose from when installing. That would be less confusing.
There will always be people who want to do that, and we'll continue to cater to them by having the wider package set remain available (as well as the spins process so people who care enough can build new install-and-deployment media).
That seems a really strange way of doing things, and it makes me think it`s overly complicated and wastes a lot of effort in that so many different distributions have to be created instead of just making one distribution that lets users pick what they want. And what if I pick "product A", whatever that might be, because I need Z, and then I find that I need X from "product B" as well. I can`t have it because I made the wrong pick and need to create another distribution myself to get it because there is no Fedora distribution that has both X and Z?
No wonder that people don`t use Fedora ...
getting user input into Fedora. How could that be done better? Surveys? More user testing? An active "User Feedback SIG"?
I think that a mailing list like this one can provide a lot of input
Mailing lists are a *start*, but you also have to recognize that you're dealing with a self-selected set of responders. [...] This is traditionally a set of people who have established their own ways of working around (and sometimes mentally blocking) some of the more painful parts of the Fedora experience.
And this set would be quite different from what Fedora thinks it`s user base[1] is, or should be.
What actually is the user base?
[1]: http://fedoraproject.org/wiki/User_base
Put another way, limiting our source of input to the mailing list would be fundamentally equivalent to devising a project's budget estimate by only talking to the engineers. By not talking to marketing, quality-assurance, capital expenditures, facilities, etc., you'd come up with an inaccurate view.
I`m not saying the input should be limited to what can be found on mailing lists like this one.
If what users think is supposed to be relevant, and if the assumed user base is to be considered, you`d have to get input from people who never have used Fedora just as much as from the people using it.
Since we seem to have established that what users think is irrelevant unless they make it happen themselves, you remain pretty limited to the sub-set of the self-selected responders you already have --- i. e. the people that make something happen --- and there is nothing wrong with that because you say it is supposed to be like this and will never be any different.
That partly answers your question whether the Fedora project is positioned to lead the advancement of FOSS, and the answer is "no".
Perhaps the way out of this is to change your thinking. Until then, no more input from users is required, and Fedora has to remain limited to what the people who make something happen might come up with.
How do you currently find out what users want?
In my experience, we almost never find out what users want. We *often* hear after the fact what they don't like, but we rarely if ever hear recommendations ahead of time.
Why is that? Because users aren`t aware of what`s coming up, and those who are don`t have a way to be listened to?
Furthermore, users are often REALLY BAD at knowing what they actually want. [...] The moral here is that what they asked for and what they needed were not the same thing.
Even matter of factly so :) You had to figure it out together with them. Bug reports are not very well suited for that; unless they are feature requests, they don`t deal with what users want, but with bugs.
When they are feature requests, you right away doubt that what the users want might be what they need, and feature requests in a bug-tracking system are not very well suited to figure it out together with them, are they?
At last, the users might be rather happy with what they have, and in that case, the problem that they might not need what they want doesn`t apply. Trying to give them something else instead is bound to meet resistance. They just picked something, and it`s fine. Change it, and/or try to give them something else, and you make them unhappy. (That seems to happen *often*, as you say?)
You could say that Fedora has the wrong users, and that Fedora.next may need to reconsider the assumed user base.
Suppose I`d say that Fedora is an experimental Linux distribution not designed for "ordinary users" but for package managers. Fedora.next is an attempt to solidify this. Would I be too wrong?
Matthew Miller mattdm@fedoraproject.org writes:
On Mon, Mar 24, 2014 at 08:19:52PM +0100, lee wrote:
/usr belongs on it`s own partition.
As if no one has ever said that before, and as if it convinced even one thinking person to change their mind.
Thinking persons do not need to change their minds about it because they realise that being able to have /usr on it`s own partition is a good thing.
It's important to realize that you *can* have a separate /usr -- it just really needs to be available at boot time.
The F17 installer wouldn`t let me have it.
That means you can have separate mount options, filesystems, partition constraints, or whatever. It just doesn't work anymore to have it on a network share or (if anyone ever did this!) removable media added after initial boot.
But it works when you plug it in before booting?
I used the network share case in the mid 1990s, when we were trying to cram Irix 6 onto 800MB workstation drives. These days, that's not really an issue. (And, hey, you can fit minimal Fedora in that space!) It might be neat for some special cases, but I hope we can all agree that it *is* a special case (and that Fedora isn't necessarily the right thing to cover all special cases).
It`s probably not totally impossible to do it, or is it?
On Tue, 2014-03-25 at 13:09 +0100, lee wrote:
That means you can have separate mount options, filesystems,
partition
constraints, or whatever. It just doesn't work anymore to have it on
a
network share or (if anyone ever did this!) removable media added after initial boot.
But it works when you plug it in before booting?
It's not enough to plug it in. It has to be mounted, meaning the network has to be up quite early in the boot procedure (details left as an exercise for the reader). I think that's was the point.
poc
On 22 March 2014 20:05, Chris Murphy lists@colorremedies.com wrote:
Ten years of failure installing Fedora. You should make your own t-shirt.
Yeah, go me.
Actually, no, I don't consider that my job is to do the developers' work for them. My job, the one I was being paid for, was to advise people interested in trying Linux what distros to look at.
E.g.
http://www.theinquirer.net/inquirer/news/1003013/inquirer-guide-free-operati...
Or a few years later this:
http://www.theregister.co.uk/Print/2010/06/21/reg_linux_guide_1/
You will note that my verdict is not positive concerning Fedora and I tend to suggest people use Ubuntu or failing that CentOS.
In those ten years, you're still unfamiliar with the installer's bug reporting mechanism?
Yeah, oddly, I remain unfamiliar with the components of an OS that I can't get to install. Funny, that.
In lieu of the fact you have a 0% install success rate installing Fedora on baremetal, aren't you suspicious? Curious?
Some of all of them, sure, but my time is limited and my job in the exercise was to find the best tool for various jobs. If you're comparing a dozen cars, you don't spend a week trying to find out why one of them won't start - you just note that it didn't complete the test and move on.
Have you inquired with anyone about how to get more information from the installer to possibly find out why you've had a decade of install failures, on just Fedora? Are you unfamiliar with the installer's rather substantial logging feature, always enabled, with easily recovered logs? Have you looked at them, or had anyone else look at them? What was the cause of the failure?
Nope. See above. "Failed to complete tests". Actually, "failed to even start tests".
-- Liam Proven * Profile: http://lproven.livejournal.com/profile Email: lproven@cix.co.uk * GMail/G+/Twitter/Flickr/Facebook: lproven MSN: lproven@hotmail.com * Skype/AIM/Yahoo/LinkedIn: liamproven Tel: +44 20-8685-0498 * Cell: +44 7939-087884
On 24 March 2014 16:11, Ralf Corsepius rc040203@freenet.de wrote:
These aren't old-fashioned, these are technically out-dated. Makes a huge difference!
I am not sure it does. I would tend to consider them as 2 sides of the same coin.
The fact you haven't encountered it doesn't mean there are no use cases.
Agreed. That was my point to Mr Murphy. :¬)
Just think about non-desktop HW (e.g. phones, tablets, routers, switches, NASes), which usually are equipped with different types of memory, being used for different purposes ("Linux as firmware").
OK, true, but does Fedora run on any of these? Or even RHEL? I don't think so.
Definitely. Such setups are not uncommon on servers and are even sold by big brands. e.g. HP.
E.g. the HP ProLiant N36/40/54L - These are equipped with a built-in usb-2 socket, designated to take an USB-stick to boot the OS from.
I have just been robbed of £92 by a bad eBay vendor trying to buy one, actually, or I would have known this. It was my plan to boot if off a hard disk, though.
Similar setups also aren't uncommon on HTPCs, NAS-boxes and similar boxes where "non-data partition"-filesystem performance is not of much importance.
Hmm. OK. I will take your word for it; I have some experience with such things and [a] I have /very/ rarely seen strange partitioning layouts and [b] I've almost never seen Fedora on them.
But is it used, is it really accessed? I guess no.
Yes, really, it is. Most commonly 1 LCD/keyboard-with-trackball on a KVM in each rack.
Also think about NASes or boxes being used as routers. No need for graphics on them.
No need, no, but it is built into even low-end CPUs these days!
No need to do so. RH has implemented facts which have rendered this discussion moot. IMO, some hidden cabal at RH had decided to pick the ancient (> 20 years old) idea to abandon separate partions for /usr and / and to sell it as "revolutionary novelty", instead of shooting it down such proposals as "Windows way of thinking", as it has been done for 20 years before :)
Well, actually, I am generally in favour of people reconsidering tradition and looking for better ways to do things, as a rule.
One of my favourite experimental distros is GoboLinux, which completely gets rid of the traditional Unix hierarchy; all packages go in their own subtrees, with plain English names, and the traditional Unix directories full of huge collections of libraries from hundreds of different programs is faked with symlinks. It is very clever and *much* more accessible and comprehensible than traditional Unix systems.
I also really like projects like OSv, to design dedicated OSes just to run inside VMs serving single tasks, reducing the huge duplication involved in current whole-system-emulation x86 virtualisation.
Throwing out tradition is good sometimes. And yes, that will sometimes break old workflow and practice. That's OK if there is a clear benefit.
On Tue, Mar 25, 2014 at 01:09:12PM +0100, lee wrote:
It's important to realize that you *can* have a separate /usr -- it just really needs to be available at boot time.
The F17 installer wouldn`t let me have it.
Yeah, but F20 installer does. What was the complaint again? :)
That means you can have separate mount options, filesystems, partition constraints, or whatever. It just doesn't work anymore to have it on a network share or (if anyone ever did this!) removable media added after initial boot.
But it works when you plug it in before booting?
Presumably. I haven't tried. Of course, you'll need whatever is necessary to load that disk in the initramfs.
I used the network share case in the mid 1990s, when we were trying to cram Irix 6 onto 800MB workstation drives. These days, that's not really an issue. (And, hey, you can fit minimal Fedora in that space!) It might be neat for some special cases, but I hope we can all agree that it *is* a special case (and that Fedora isn't necessarily the right thing to cover all special cases).
It`s probably not totally impossible to do it, or is it?
Probably not, but you might need to do some special work (including possible changes to "normal" Fedora) and that's probably okay.
On 03/25/2014 02:08 PM, Liam Proven wrote:
On 24 March 2014 16:11, Ralf Corsepius rc040203@freenet.de wrote:
These aren't old-fashioned, these are technically out-dated. Makes a huge difference!
I am not sure it does. I would tend to consider them as 2 sides of the same coin.
Have a look at "fashion" vs. "progress".
Somebody wearing 19th century clothes is old-fashioned, somebody driving a coach is using out-dated technology.
This of course doesn't mean there are no valid use cases for 19th century clothes nor using coaches.
Just think about non-desktop HW (e.g. phones, tablets, routers, switches, NASes), which usually are equipped with different types of memory, being used for different purposes ("Linux as firmware").
OK, true, but does Fedora run on any of these? Or even RHEL? I don't think so.
Well, if they are based on architectures Fedora supports, Fedora theoretically could be made working on them.
ATM, this only applies to PC-based HW, such as some NASes/routers/HTPCs etc.
Definitely. Such setups are not uncommon on servers and are even sold by big brands. e.g. HP.
E.g. the HP ProLiant N36/40/54L - These are equipped with a built-in usb-2 socket, designated to take an USB-stick to boot the OS from.
I have just been robbed of £92 by a bad eBay vendor trying to buy one, actually, or I would have known this. It was my plan to boot if off a hard disk, though.
I own an N40L. It currently is running Fedora and basically serves as my (home-) office's network's main multi-purpose server. When it was new, due to lack of a spare HD, I was using an 8GB USB-stick as system drive. I was facing malfunctions after a couple of months of operation. I don't know the real cause, but am inclined Fedora may have caused wear of the USB-stick :=)
But is it used, is it really accessed? I guess no.
Yes, really, it is. Most commonly 1 LCD/keyboard-with-trackball on a KVM in each rack.
Also think about NASes or boxes being used as routers. No need for graphics on them.
No need, no, but it is built into even low-end CPUs these days!
Right .. but not into all, nor do all Mobos support CPU-built-in graphics.
Ralf
On Tue, 2014-03-25 at 07:13 -0400, Matthew Miller wrote:
On Mon, Mar 24, 2014 at 08:19:52PM +0100, lee wrote:
/usr belongs on it`s own partition.
As if no one has ever said that before, and as if it convinced even one thinking person to change their mind.
Thinking persons do not need to change their minds about it because they realise that being able to have /usr on it`s own partition is a good thing.
It's important to realize that you *can* have a separate /usr -- it just really needs to be available at boot time. That means you can have separate mount options, filesystems, partition constraints, or whatever. It just doesn't work anymore to have it on a network share or (if anyone ever did this!) removable media added after initial boot.
I used the network share case in the mid 1990s, when we were trying to cram Irix 6 onto 800MB workstation drives. These days, that's not really an issue. (And, hey, you can fit minimal Fedora in that space!) It might be neat for some special cases, but I hope we can all agree that it *is* a special case (and that Fedora isn't necessarily the right thing to cover all special cases).
-- Matthew Miller -- Fedora Project -- mattdm@fedoraproject.org
But in the modern business environment, users log in from multiple places. How does that work if the user directory is local?
On Mar 25, 2014, at 2:41 AM, lee lee@yun.yagibdah.de wrote:
Chris Murphy lists@colorremedies.com writes:
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
Please post the bugzilla URL.
I didn`t make one, I was busy trying to install. You need to get used to that not every problem is or can be made into a bug report, especially not one that would be in any way useful.
What I ought to do is not QA Manual Partitioning anymore and just let the people who actually think they need or want it do all the testing and bug reporting for it and let it become whatever it becomes.
Most of the size reporting problems like this are non-contiguous sections of free space being added up and reported as Available space; but the request is for a partition size greater than the largest contiguously available space.
Maybe it begins with the installer messing together all the disks in some weird way rather than to treat them separately and just let you partition them the way you want to. IIRC, there wasn`t even a way to tell it which partition to put where.
It is possible although the UI isn't obvious. You click on the mount point in question, and then there is this 3rd or 4th button under the mount points section that looks like a wrench and screwdriver (?) that you click. And that brings up a dialog where you choose which drive that mount point's underlying partition appears on.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
It isn't going to get better complaining about it on this list. Do you have bugzilla IDs, and if so post them. If not, then how do you expect the behavior to get any better? Magic?
The makers of the installer can always look into this list and see what ppl say about the installer and learn from that.
No they will not do this, and it's inappropriate to even suggest it. That you don't get that simply means you're ignorant of how the process works.
Bug reports are not suited for this, and complaining that ppl don`t make enough of them doesn`t get you anywhere.
Filing a bug report is the process. That's it. It works this way for everything: gnome, kde, and even commercial projects do it this way. They do not have developers hanging out in user forums ever. Sometimes QA people hang out in user forums.
Or, since you keep insisting on bug reports, why don`t you go ahead and put together a list of URLs to the list archive pointing to posts about the installer to compile a combined bug report?
Why don't you go ahead and send me 4-5 bitcoins and I'll think about it?
This is how I know your problem isn't really serious, because it's so unimportant to you, you won't lift a finger to contribute to any improvement. Why should I help you since you won't even help yourself?
How much attention and fixing do you think that would get? It`ll probably be closed with WONTFIX, one reason being that what is said here doesn`t refers much more to installers from F17 to F19.
Perhaps the installer of F20 has be re-designed from scratch, at least when it comes to partitioning, and now works fine.
The installer is redesigned as of Fedora 18.
you just do country and keyboard setup --- which is missing in Fedoras installer, there was no way to tell it that I have a German keyboard ---
Fedora 18, 19 and 20 have a keyboard spoke in the installer which is how you tell it you want to use a German keyboard layout.
Whatever they might have, I tried several times (because I had to start over many times because it refused to do the partitioning) to tell the installer of F19 that I have a German keyboard, and there was no way.
OK well I just tried it on Fedora 19 Netinstall and Fedora 20 Live Desktop and I could easily choose a German keyboard, so I don't know what you're doing wrong. It's right there under Localization in the Keyboard option on the main menu right after choosing language.
I don`t even know what you mean by "spoke". You boot the life system, search for "install" to find the installer, then you get an icon and start the installer, and pretty soon you get stuck with trying to do the partitioning.
Hub = the main menu that comes after language selection. On the hub are "spokes" which are paths for various things like "Time & Date" and "Keyboard" and "Installation Destination".
Chris Murphy
I`m not an expert with Fedoras installers in any way. This is simply my "user experience". Maybe the "user experience" the installer provides should be different.
Your user experience is important, but just because you think it should be one way doesn't mean the community or developers echo that sentiment.
A non-gui installation is not something that the majority of users will choose so it's not apparent, but if you want that method, here you go: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/ ch-guimode-x86.html#idm219166212128
Says who? And why not give the users that choice instead of hiding it?
The option or choice is given; why not make it more obvious? Because it would most likely pose more of a distraction to the target audience through clutter than it would ever get used; hardcore users that might favor a text installer are not the target audience anymore.
Keyboard configuration is not missing... it's one of the main hub options: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/ install-hub-x86.html
That one looks rather unfamiliar. Was that in the F17 or F19 installers?
F18/19/20
The computer industry knows that more and more people want installations to be less scary and faster. This trend has been seen in the evolution of the Windows installer, MacOS, most Linux distros, and even iOS or Android. There are going to be some Linux distros that don't embrace this, you mentioned Debian and I'm sure slackware and Arch won't either, but in the end it's all about attracting more users to the product. The choices are there, but us hardcore users just have to look more since we're the minority.
Faster, ok, that mainly depends on what the bandwidth of your internet connection is and what the server side delivers. The only two scary things are the possibility that the network device doesn`t work (like Debian missing modules for some) and, far worse, potentially loosing your data.
Not really. Fedora has three installation images: Live OS, Full DVD, and Network. The only installation image that would be affected by internet connectivity is the Network Install image, because it downloads all packages in lieu of bundling packages. The Full DVD does allow the user to setup third part repos, but that usually is a minimal hit at worst.
Having more choices doesn`t make installing any slower or any more scary. It makes it easier. Not having choices makes installing much slower and more scary.
I disagree; if a user is presented with the following filesystem choices, btrfs, ext2, ext3, ext4, JFS, reiser4, reiserFS, and ZFS, and each is presented equally with a single paragraph describing its benefits, unless the user has prior knowledge about what is the best choice for the intended installation goal, they're most likely going to spend a great deal of time reading each paragraph. If you simplify the choices to 4 instead of 8, the user has less paragraphs to read and can make a decision faster.
On 03/25/2014 12:10 PM, Powell, Michael wrote:
I disagree; if a user is presented with the following filesystem choices, btrfs, ext2, ext3, ext4, JFS, reiser4, reiserFS, and ZFS, and each is presented equally with a single paragraph describing its benefits, unless the user has prior knowledge about what is the best choice for the intended installation goal, they're most likely going to spend a great deal of time reading each paragraph. If you simplify the choices to 4 instead of 8, the user has less paragraphs to read and can make a decision faster.
Agreed. However, it might be a good compromise to list only the most common selections, with a button marked Advanced Options (There's probably a better label, but if so, I can't think of it right now.) containing the rest. Beginners will, most likely, avoid that, but those who need/want the more specialized file systems will still have access to them.
On Tue, Mar 25, 2014 at 09:50:10AM -0700, Howard Howell wrote:
It's important to realize that you *can* have a separate /usr -- it just really needs to be available at boot time. That means you can have separate mount options, filesystems, partition constraints, or whatever. It just doesn't work anymore to have it on a network share or (if anyone ever did this!) removable media added after initial boot.
But in the modern business environment, users log in from multiple places. How does that work if the user directory is local?
On modern Linux/Unix, the "/usr" directory holds system binaries and libraries -- it is not the user directory. On Fedora (and most Linux systems), that is "/home". And there's no problem sharing that over the network.
On 25/03/14 22:29, lee wrote:
Tim ignored_mailbox@yahoo.com.au writes:
I reckon it's the case for most OSs that /most/ users don't really care much about what they're using, nor how it works. The large number of clueless people using computers would seem to be evidence of that.
The number of clueless people is also large without computers. Computers might only make that number larger.
It's a lesser number of people that have concerns about their OS. Maybe that's why developers are less concerned about public opinions.
Or maybe people don`t bother to have or bring about concerns because the developers don`t care.
And on it goes! Fedora and Linux in general is what it is. We are fantastically lucky that developers and testers donate vast knowledge, time and resources. We have choice, so many variations.
If I were a developer thinking about improvement, speed, constant changes in architecture and incessant increasingly aggressive system attacks, which you and I are so sheltered from, I would not give a whit about what users think. There are others who handle that part of the development. Please show some gratitude for those people who devote their all to giving us a free unfettered operating system. I, and my family are devoted Linux users for way over ten years and grateful for the opportunity. We don't care what it looks like or how it operates, we adapt and continue working as do hundreds of thousands of users.
Roger
On 03/25/2014 07:09 PM, Roger wrote:
And on it goes! Fedora and Linux in general is what it is. We are fantastically lucky that developers and testers donate vast knowledge, time and resources. We have choice, so many variations.
If I were a developer thinking about improvement, speed, constant changes in architecture and incessant increasingly aggressive system attacks, which you and I are so sheltered from, I would not give a whit about what users think. There are others who handle that part of the development. Please show some gratitude for those people who devote their all to giving us a free unfettered operating system. I, and my family are devoted Linux users for way over ten years and grateful for the opportunity. We don't care what it looks like or how it operates, we adapt and continue working as do hundreds of thousands of users.
Roger
+1 very well said Roger, thanks!
On 2014-03-23 13:49, Powell, Michael wrote:
Then they need to turn around at once! Leaving users in the dark about what`s going on makes things more difficult for them. Taking away choices limits the use of the software to the point where it eventually becomes unusable.
Good design, clarity, good documentation and letting the user know what choices there are are required when you want to achieve ease of use.
I don't think the installer is as bad as most people on this list are complaining about; it's certainly not leaving any users in the dark. A large majority of the information from the older installer is still there, it's just up to the user to seek it out.
It is leaving some people in the dark. I have use Red Hat/Fedora since RH4 and have found that the recent changes to the installation routine is a real pain if you don't want LVM and the default install.
It took multiple attempts to find the path to manually partition the drive the way I wanted.
I have now installed F20 on 4 different machines. 3 from clean installs and one with fedup. Two with Windows dual boot.
I have shown people at work how to manually partition using F18 and up. One person installed F14 on a machine because it was what he had at the time and new how to manually partition the drive.
The problem is having the user seek it out. There could be a simple option for the default partitioning or custom in a much clearer instruction. I must have cancelled multiple installs with the new installer trying to find the correct install method.
Robin
On 2014-03-25 02:22, lee wrote:
"Powell, Michael" Michael_Powell@mentor.com writes:
It doesn`t give you choices. It leaves you in the dark about that it is somehow possible to use an non-gui installer and to do a minimal install. It leaves you in the dark about what exactly happens when you do the partitioning and with trying to figure out how get the partitioning you want.
As I said before, the entire installer was rebuilt to promote a more laid back approach. The user can choose the order they wish to customize / experience the GUI installation instead of being forced down a specific path. There might be a couple mandatories, but for the most part it's all about choice.
Which installer are you referring to? As to Fedora, I have only used the ones of F17 and of F19, and none of them gave me any choices. You can start the installer and it doesn`t even let you do the partitioning you want, which is the only choice you are getting.
I`m not an expert with Fedoras installers in any way. This is simply my "user experience". Maybe the "user experience" the installer provides should be different.
It is hidden. I do use it.
If you were trying the live disk, then I believe that the manual partitioning isn't there. I was told on this list to use the full DVD. Yes, I use DVD as an install medium. There is a reason.
A non-gui installation is not something that the majority of users will choose so it's not apparent, but if you want that method, here you go: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/ch-gui...
Says who? And why not give the users that choice instead of hiding it?
I agree. It should be easier to find.
On 2014-03-25 13:26, Joe Zeff wrote:
On 03/25/2014 12:10 PM, Powell, Michael wrote:
I disagree; if a user is presented with the following filesystem choices, btrfs, ext2, ext3, ext4, JFS, reiser4, reiserFS, and ZFS, and each is presented equally with a single paragraph describing its benefits, unless the user has prior knowledge about what is the best choice for the intended installation goal, they're most likely going to spend a great deal of time reading each paragraph. If you simplify the choices to 4 instead of 8, the user has less paragraphs to read and can make a decision faster.
Agreed. However, it might be a good compromise to list only the most common selections, with a button marked Advanced Options (There's probably a better label, but if so, I can't think of it right now.) containing the rest. Beginners will, most likely, avoid that, but those who need/want the more specialized file systems will still have access to them.
This is something that I totally agree with. And an final commit button. I have had to abort installs because I missed some setting and couldn't remember where "That Damn Setting" was and clicked the wrong button.
On 2014-03-24 08:25, Liam Proven wrote:
On 23 March 2014 21:56, lee lee@yun.yagibdah.de wrote:
Installing on a laptop requires encrypted partitions. They can be stolen too easily.
I have never ever used this and never expect or plan to. I suggest that your blanket statement is too sweeping.
This is good but I agree with Lee, encryption is necessary on a laptop, a business one especially. Lack of encryption could be a court case.
Nowadays you may have SSDs which supposedly last longer when not written much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
Machines come with dozens of gigs of RAM now. I'm not sure there's much argument for swap at all, and personally, I use tmpfs for better performance and a self-cleaning /tmp tree.
New machines come with dozens of gigs of ram. Try getting a laptop with that much ram. tmpfs is a great idea for privacy but it is a pain on a lower end machine. Found that out the hard way.
And it`s not too difficult. The installer doesn`t need to do the partitioning, the user does it. The installer only needs to give the user a good tool to do the partitioning the user wants and let them use it. Good tools to do partitioning are already available, and the installer doesn`t need to re-invent the wheel in that.
I mostly agree, but bear in mind that the installer must cope with both experts and novices. That's a tough call.
I'd say it's /too/ simplified at the moment, though, and I think you might agree...?
And that needs to be an option that is clear to the person doing the install. Simple choice. Easy or custom. Not clicking Done and then getting to choose the partitioning options.
Perhaps it even shouldn`t. Why force the user to learn how to use yet another partitioning tool they even rarely use unless they install Fedora all the time? Why not give them a choice, like either cfdisk or parted, then tell the installer what to do with each partition and let them switch between these until they are done --- or let the installer do whatever partitioning it wants, which means that all existing data on the disks will be deleted.
Mostly, I would agree, actually.
On 2014-03-22 10:38, lee wrote:
Matthew Miller mattdm@fedoraproject.org writes:
On Fri, Mar 21, 2014 at 04:19:05PM +0100, lee wrote:
And on top of that, what is the Fedora-way of replacing gnome --- which I find totally useless --- with fvwm, which perfectly does what I want?
It sounds like you want to do a minimal install and then add up from that.
Yes, that`s what I always did with Debian.
I do a minimal install with the Fedora DVD (haven't tried f20 and live disk yet). Then update and install packages after. No need to install a bunch of packages from the DVD and then have to do 1200 updates.
Seems faster. Just need to setup a repository at my home to save bandwidth. :)
I think you will benefit from this effort in that the minimal install will be better defined and curated.
That would be nice --- I wouldn`t even have thought that there is one if I hadn`t read on this list that there is one, somewhere. I still don`t know how I would start with a minimal install, though.
When you get the installer and boot it, you get a working system. That`s a good way to go because otherwise you need a second computer around when installing to look up things. But where is the minimal install, and what when you don`t get a GUI?
Using the DVD, just go into the packages and deselect all the packages that you don't need for a minimal install. I like yumex and then use the group feature to install other applications. I use KDE, not gnome.
It`s only one example, and you can figure out how to do it. But the point with this is that Fedora lacks flexibility. You get what you get and then have to go through a lengthy process of getting rid of stuff and of somehow getting to work what you need, like fvwm.
Sure, I would agree that this isn't a strong suit, particularly with the all-or-nothing way RPM dependencies currently work.
Yes, that is one of the things I don`t like. It forces you to install stuff you never need.
RPM is better than it used to be if the maintainers set their spec files properly and only include what is needed. Now a lot of libraries and parts are removed from the main packages.
YUM is the way to go as it checks the requirements and installs only what is needed, most times.
On a side note: The installer sucks, just try to do one of the most basic and important things with it: Partitioning.
Saying something "sucks" isn't very helpful. Not only is it needlessly negative, it is intangible. Name a real problem and we can talk about it.
There have been a few threads about it on this list. The major problem is partitioning.
If you use the correct method, you can partition your drives. I have done it on most installs I have done.
The last time I used the installer, it was the one F19 comes with. It was impossible to get the partitioning I wanted, so I had to partition otherwise. Then it was nearly impossible to make the installer actually use the partitions the way I wanted. And since the buttons the installer uses are weird and misleading, you never really know what you`re doing. I had to try over and over again to figure out how to somehow make it use the existing partitions. If I had used it on a computer that had data on the disks I wanted to keep, I`d have had to physically unplug them to make sure the data doesn`t get lost --- and that isn`t always possible.
I agree that the buttons are misleading. Click Done to get to custom partitioning is strange. I have to do some testing to learn more. I don't have a decent machine for setting up and running VM's.
So some simple partitioning that would be done with the Debian installer within ten minutes took three hours. The Fedora installer does what it want`s, not what the user wants, and it leaves the user in the dark about what is actually going to happen.
On top of that, it doesn`t offer any choices. You cannot pick any package at all.
This wrong. You can pick packages if you use the full DVD. I am not sure about the Live disks though. I pick a minimal install on the DVD.
I was at a large cloud conference a while ago, and almost nobody was using Fedora, and so I asked people why they chose the distribution they are building their stuff on, and why they didn’t choose Fedora. Almost universally, the response wasn’t “What I am using is great!” — it was “Oh, I don’t care. I just picked this, and that’s what I’m using and it’s fine.”
I`m not one of these people. Thinking like that, they don`t need a Linux distribution; they can as well use Windoze or Macos.
Yet these people were absolutely running Linux. Just not ours.
Hm. Maybe that`s how people are.
Think of other things they use: How many times do they pick something because it`s great? It involves more work to do that because you need to find out what is available, what the differences are and what makes something great for you in particular. So they don`t bother and use something which is fine.
Look at http://fedoraproject.org/ --- it says this and that and doesn`t say or show that Fedora is great or awesome and why.
This is an interesting aspect. It fits in with my father. He started using Linux for a project he was involved with. It was Centos based and it just worked. He started experimenting and I got him to try Fedora. Last Linux he played with on his desktop was Mint which he loved.
One of our system admins moved to Ubuntu from Fedora because of broken apps and almost impossible upgrade path at the time. I tried Fedup and won't use it on my present machines. I can do a full install and configuration much quicker.
On 2014-03-23 10:19, Tom Horsley wrote:
On Sun, 23 Mar 2014 16:35:20 +0100 lee wrote:
One of the weaknesses of Fedora, in my view, is the apparent lack of interest in what users actually want or need.
+1
The distro that actually seems to care (at the moment) about users is Linux Mint (though I have no doubt that, like all open source projects, it will eventually be taken over by a cadre of "helpful" zealots who will yank the steering wheel in different direction because stupid users don't realize what they actually need :-).
I use Fedora primarily because it is the best distro to use as an early warning system to find out what nonsense will land in RHEL in the future, and at work we need our software to run on RHEL, so Fedora is sort of like the Canary in the coal mine.
I think this is why my father went to Mint over Fedora and Centos when he did.
I like the analogy of the Canary. It fits. Like F20, many broken apps out of the box. I need to get an clean install and test so I can report the bugs.
On 2014-03-25 05:29, lee wrote:
Tim ignored_mailbox@yahoo.com.au writes:
I reckon it's the case for most OSs that /most/ users don't really care much about what they're using, nor how it works. The large number of clueless people using computers would seem to be evidence of that.
The number of clueless people is also large without computers. Computers might only make that number larger.
True on both accounts. Just like a discussion on a different list I am on.
It's a lesser number of people that have concerns about their OS. Maybe that's why developers are less concerned about public opinions.
Or maybe people don`t bother to have or bring about concerns because the developers don`t care.
I don't agree. In many cases it is lack of feedback from users. I know many developers that I have emailed about problems and they do respond. Now don't confuse package maintainers with developers. Two different things.
If I have a complaint in this aspect about Fedora is about major upgrades to applications and the policy of not feeding those upgrades to present users. VLC is an example. Fixes in upstream VLC to fix the save screen shot as a PNG but not fed to users of F19. Now this is an rpmfusion issue but can be said for Fedora maintainers as well. I remember an issue with pam_mount a few years ago.
I would really like to see Fedora move to a constant flux where applications are developed and then pushed out as updates which are actually upgrades. Get to a point of not having a F19 or F20 but just Fedora. You move from 19 to 20 to 21 through the daily updates instead of a major upgrade.
On 03/26/2014 06:45 AM, Robin Laing wrote:
On 2014-03-24 08:25, Liam Proven wrote:
On 23 March 2014 21:56, lee lee@yun.yagibdah.de wrote:
Nowadays you may have SSDs which supposedly last longer when not written much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
Machines come with dozens of gigs of RAM now. I'm not sure there's much argument for swap at all, and personally, I use tmpfs for better performance and a self-cleaning /tmp tree.
New machines come with dozens of gigs of ram.
Exactly, but old one don't and old ones often can not even be upgraded.
That said, consider many so-far-WinXP users currently are trying to migrate to other OSes. I can't deny finding it poor, Fedora is not an option to many of them, because of Fedora's memory requirements [1].
Ralf
[1] From my experience, F20 can be made runable on machines with 512MB RAM, but is hardly installable because the installer requires somewhat less than 1GB RAM.
On Wed, Mar 26, 2014 at 12:14:44AM -0600, Robin Laing wrote:
I would really like to see Fedora move to a constant flux where applications are developed and then pushed out as updates which are actually upgrades.
It is Fedora policy to not do that:
http://fedoraproject.org/wiki/Updates_Policy#Philosophy
Get to a point of not having a F19 or F20 but just Fedora. You move from 19 to 20 to 21 through the daily updates instead of a major upgrade.
You are talking about a rolling release, this topic has been discussed to death and rejected.
http://www.phoronix.com/scan.php?page=news_item&px=MTIyMDg
Cheers,
On Wed, Mar 26, 2014 at 12:14:44AM -0600, Robin Laing wrote:
I would really like to see Fedora move to a constant flux where applications are developed and then pushed out as updates which are actually upgrades. Get to a point of not having a F19 or F20 but just Fedora. You move from 19 to 20 to 21 through the daily updates instead of a major upgrade.
You might want to try running Rawhide, our development branch. It works exactly that way. Sometimes there are issues, but... that's thr price you pay for that.
On 03/26/2014 01:32 AM, Robin Laing wrote:
I`m not an expert with Fedoras installers in any way. This is simply my "user experience". Maybe the "user experience" the installer provides should be different.
It is hidden. I do use it.
If you were trying the live disk, then I believe that the manual partitioning isn't there. I was told on this list to use the full DVD. Yes, I use DVD as an install medium. There is a reason.
I came from Debian recently with Fedora 20. I was told to use the net install ISO, I did. I have windows and another linux distro installed and used manual partitioning without a problem. I have 2 drives and used my existing /home on drive A and installed Fedora on drive B. works fine.
Chris Murphy lists@colorremedies.com writes:
On Mar 25, 2014, at 2:41 AM, lee lee@yun.yagibdah.de wrote:
Maybe it begins with the installer messing together all the disks in some weird way rather than to treat them separately and just let you partition them the way you want to. IIRC, there wasn`t even a way to tell it which partition to put where.
It is possible although the UI isn't obvious. You click on the mount point in question, and then there is this 3rd or 4th button under the mount points section that looks like a wrench and screwdriver (?) that you click. And that brings up a dialog where you choose which drive that mount point's underlying partition appears on.
Clicking on unknown icons like that is a scary thing to do when you can expect that it potentially deletes your data.
I`m not a fan of icons. I`m not that visually orientated that I would remember them or that I could somehow guess from a, often tiny and difficult to figure out picture, what something might be supposed to do.
Anyway, messing the disks together contradicts what the user is doing. You don`t create partitions in thin air but on particular disks or volumes.
It isn't going to get better complaining about it on this list. Do you have bugzilla IDs, and if so post them. If not, then how do you expect the behavior to get any better? Magic?
The makers of the installer can always look into this list and see what ppl say about the installer and learn from that.
No they will not do this, and it's inappropriate to even suggest it. That you don't get that simply means you're ignorant of how the process works.
Apparently they don`t do it, and that may very well be where some of the impression that the makers of the distribution are far away from the users and don`t care what they think comes from.
Why would it be inappropriate to suggest that they take a look? Would you rather copy all relevant posts from this mailing list into a bug report?
This discussion is about Fedora.next, which seems to be some sort of effort to figure out what Fedora needs to or should do in the future. When suggestions towards that are considered inappropriate, why are we having this discussion?
Bug reports are not suited for this, and complaining that ppl don`t make enough of them doesn`t get you anywhere.
Filing a bug report is the process. That's it. It works this way for everything: gnome, kde, and even commercial projects do it this way. They do not have developers hanging out in user forums ever. Sometimes QA people hang out in user forums.
And Fedora cannot do any better because?
Or, since you keep insisting on bug reports, why don`t you go ahead and put together a list of URLs to the list archive pointing to posts about the installer to compile a combined bug report?
Why don't you go ahead and send me 4-5 bitcoins and I'll think about it?
What are bitcoins?
This is how I know your problem isn't really serious, because it's so unimportant to you, you won't lift a finger to contribute to any improvement. Why should I help you since you won't even help yourself?
Sure, that`s why I`m taking part in this discussion. But as you wish. I`m done with this discussion, it`s obviously pointless.
Roger wrote:
If I were a developer thinking about improvement, speed, constant changes in architecture and incessant increasingly aggressive system attacks, which you and I are so sheltered from, I would not give a whit about what users think. There are others who handle that part of the development.
Who are these others?
On Tue, 25 Mar 2014, Powell, Michael wrote: [note >>>]
A non-gui installation is not something that the majority of users will choose so it's not apparent, but if you want that method, here you go: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/ ch-guimode-x86.html#idm219166212128
Says who? And why not give the users that choice instead of hiding it?
The option or choice is given; why not make it more obvious? Because it would most likely pose more of a distraction to the target audience through clutter than it would ever get used; hardcore users that might favor a text installer are not the target audience anymore.
I disagree; if a user is presented with the following filesystem choices, btrfs, ext2, ext3, ext4, JFS, reiser4, reiserFS, and ZFS, and each is presented equally with a single paragraph describing its benefits, unless the user has prior knowledge about what is the best choice for the intended installation goal, they're most likely going to spend a great deal of time reading each paragraph. If you simplify the choices to 4 instead of 8, the user has less paragraphs to read and can make a decision faster.
IIRC the last time I did an install, there was a choice labeled something like "If you don't know what you need, this is probably it." It was the top option, but not the only one. The descriptions were one-liners, not paragraphs. Perhaps that should be the model. For filesystems, need might be too strong a word: "If you don't know what you want, this will probably work."
Matthew Miller <mattdm <at> fedoraproject.org> writes:
On Tue, Mar 25, 2014 at 09:50:10AM -0700, Howard Howell wrote:
It's important to realize that you *can* have a separate /usr -- it just really needs to be available at boot time. That means you can have separate mount options, filesystems, partition constraints, or whatever. It just doesn't work anymore to have it on a network share or (if anyone ever did this!) removable media added after initial boot.
But in the modern business environment, users log in from multiple places. How does that work if the user directory is local?
On modern Linux/Unix, the "/usr" directory holds system binaries and libraries -- it is not the user directory. On Fedora (and most Linux systems), that is "/home". And there's no problem sharing that over the network.
The funny thing is that back in the earliest days of Unix, /usr is where user directories lived. When K&R ran out of room in / for programs, they looked to for a partition that had additional space available and it was /usr. Originally programs ended up in /usr/bin simply because there wasn't room for them in /bin; not for some usage reason.
Cheers, Dave
David G. Miller wrote:
The funny thing is that back in the earliest days of Unix, /usr is where user directories lived. When K&R ran out of room in / for programs, they looked to for a partition that had additional space available and it was /usr. Originally programs ended up in /usr/bin simply because there wasn't room for them in /bin; not for some usage reason.
Yes, I recall that the first Unix system I ran, version 5 on a pdp-11/23, had two (enormous) 10MB disks, one for the kernel and the other /usr .
On 26 March 2014 22:42, Timothy Murphy gayleard@eircom.net wrote:
Yes, I recall that the first Unix system I ran, version 5 on a pdp-11/23, had two (enormous) 10MB disks, one for the kernel and the other /usr .
Wow.
Well, that's *us* told. Schooled, even.
8-)
On Wed, 2014-03-26 at 22:42 +0000, Timothy Murphy wrote:
David G. Miller wrote:
The funny thing is that back in the earliest days of Unix, /usr is where user directories lived. When K&R ran out of room in / for programs, they looked to for a partition that had additional space available and it was /usr. Originally programs ended up in /usr/bin simply because there wasn't room for them in /bin; not for some usage reason.
Yes, I recall that the first Unix system I ran, version 5 on a pdp-11/23, had two (enormous) 10MB disks, one for the kernel and the other /usr .
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
poc
Allegedly, on or about 26 March 2014, Patrick O'Callaghan sent:
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
When I was young, we didn't have no computers. We used pencil and paper, and we were grateful. ;-)
On Fri, 2014-03-28 at 02:40 +1030, Tim wrote:
Allegedly, on or about 26 March 2014, Patrick O'Callaghan sent:
That ain't nuttin'. We started with an 11/45 with 48KB of core
memory
and 2 RK05's (2.2MB removable cartridge disks).
When I was young, we didn't have no computers. We used pencil and paper, and we were grateful. ;-)
Pencil and paper? We used to *dream* about having pencil and paper!
poc
PS https://www.youtube.com/watch?v=VKHFZBUTA4k (the first of many versions of this famous sketch).
On 03/27/2014 09:10 AM, Tim issued this missive:
Allegedly, on or about 26 March 2014, Patrick O'Callaghan sent:
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
When I was young, we didn't have no computers. We used pencil and paper, and we were grateful. ;-)
And you probably ate dirt and carpet fuzz as well.
(Man, I miss Richard Jeni!) ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks@alldigital.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - When all else fails, try reading the instructions. - ----------------------------------------------------------------------
On Thu, Mar 27, 2014 at 12:43:15PM -0700, Rick Stevens wrote:
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
When I was young, we didn't have no computers. We used pencil and paper, and we were grateful. ;-)
And you probably ate dirt and carpet fuzz as well.
I think the Godwin's law of the Fedora Users mailing list is: all long-enough threads eventually get to everyone competing to be the oldest with the most esoteric use of ancient computers.
Rick Stevens wrote:
On 03/21/2014 10:30 AM, Matthew Miller issued this missive:
On Fri, Mar 21, 2014 at 10:27:02AM -0700, Joe Zeff wrote:
If you do need to use sendmail, and your ISP is blocking Port 25, it's not that hard to configure things to use a smarthost. As an example, I have my own (vanity) domain and use its mail servers, over Port 587. I also have sendmail configured to use that server and port, along with the appropriate username/password. If anybody out there needs to do the same thing, instructions are at http://www.zeff.us/SMTPAuth.txt
Absolutely. But since you need to configure it before it's useful, it's arguably actively harmful to have it running by default. That's all. No one is removing MTAs from the distro.
I agree that an MTA needs to be installed unless you're installing a "server" (and Fedora really doesn't offer that model anymore).
For quite a while, Fedora had installed an MTA and started it. By default, it was configured to only listen to localhost on port 25 so it was more-or-less innocuous. The fact that F19 stopped installing an MTA by default caught a lot of people off guard.
Sure did. And made it hard to get sendmail working usefully, IIRC both sendmail.mc and firewall were set to force you to talk only to yourself.
Unlike others, however, I find the new system logging and analysis tools cumbersome and painful to use. Having a program send an email to me if it encounters issues is FAR superior to me having to plow through the logs to see if it ran correctly or not.
In general I like logs, but that's me. That way I can go through and get what I want with perl, or grep, or sed, or whatever I need. Easier than making emails recombine for a moving overview.
Tom Horsley wrote:
On Fri, 21 Mar 2014 13:13:28 -0400 Matthew Miller wrote:
Saying something "sucks" isn't very helpful. Not only is it needlessly negative, it is intangible. Name a real problem and we can talk about it.
Every single aspect of it is a real problem. In the dictionary they have a picture of it next to the word "disaster". The whole random wheel/spoke thing is an invitation to forget to do something important. And who the hell would think that "network" would be the spoke where you set the system name? Why is the screen full of blank space and cryptic meaningless icons when it could use some of that space for valuable hints in actual text you can read?
I'd love to give detail of how badly it sucks, but just as an example, I had a machine with fc14 and fc17 installed. They shared /home, swap, and /local/appgrafix. Trying to get rid of fc17 and put in fc19 or fc20 also sharing those partitions was a long futile endeavor trying to find the "do what I say and shut up" option. Nice to protect people who don't know what they're doing, but the whole effort to overcome the protection was a waste of half a day using man pages, shared doc, wiki pages, search engines, and mailing list postings.
On the specific you do give, I'm pretty confident in saying that you're actually wrong.
And therein lies the problem. Everyone who worked on partitioning has been congratulating each other for so long you can't see any problems. The partitioning is actually absolutely impossible to use:
I won't waste time agreeing point by point, I'll just agree that anyone who didn't write this or help develop it is likely to do things the way you wanted or go elsewhere. And frankly if I wanted someone making decisions for me I would use Windows.
If there are multiple disks, it provides you only the size, model, and serial number to distinguish them, as though everyone has that memorized. Your only actually choice is to pick them all and hope you can see a clue later.
After picking which disks to partition, you are presented with a single choice: You can press the Done button even though you are not anywhere close to done, or you can say "Screw this" and install a different distro.
Should you work up the courage to press the Done button, you then reach the next layer of total obfuscation.
It names the partitions by what operating systems are installed on them. Who the hell thinks that way? And where do I look for partitions that don't have any operating system on them. Why not name them by the biggest video file that lives on the partition or some other random choice?
The adjectives "intuitive" or "useful" cannot be applied to any part of this interface. It is absolute and total junk.
Our only hope is for redhat to use this in the next RHEL and then we can probably get changes as wave after wave of enterprise users descend on redhat HQ with pitchforks and torches.
I hate to say it, but enterprise users will either use it as the great crimson chapeau thinks best, or figure out how to game the system, or go to another distro. Only the last and least desirable action will be visible to management.
Doing real installs on real machines, complex stuff with hardware and software RAID, other things already on the disk, etc, installs by a mix of people coming from Windows admin, old hands who expect the old way, and fresh from RHEL6 without any help but the docs which come with the machine.
Very few admins do a lot of installs, at peak I doubt I did more than 30 a year, and almost all servers, so few will be install gurus. If Fedora is to be the icebreaker for RHEL, it has to be better.
On 2014-03-26 00:13, Ralf Corsepius wrote:
On 03/26/2014 06:45 AM, Robin Laing wrote:
On 2014-03-24 08:25, Liam Proven wrote:
On 23 March 2014 21:56, lee lee@yun.yagibdah.de wrote:
Nowadays you may have SSDs which supposedly last longer when not written much to but mostly read from, so you might put the partitions that can be read-only on the SSDs and use magnetic disks for things like /var, /tmp, /home and swap.
Machines come with dozens of gigs of RAM now. I'm not sure there's much argument for swap at all, and personally, I use tmpfs for better performance and a self-cleaning /tmp tree.
New machines come with dozens of gigs of ram.
Exactly, but old one don't and old ones often can not even be upgraded.
That said, consider many so-far-WinXP users currently are trying to migrate to other OSes. I can't deny finding it poor, Fedora is not an option to many of them, because of Fedora's memory requirements [1].
Ralf
[1] From my experience, F20 can be made runable on machines with 512MB RAM, but is hardly installable because the installer requires somewhat less than 1GB RAM.
But that is the move forward with all OS's.
I agree, for XP users, the need for 1 or 2 gig is going to be a limit. I have a couple of machines that have 250M of ram and I had to swap around memory modules to get an install on one of them. Now they are just testing machines for drives and hardware things.
I would love to get all these Win XP users onto Linux, preferably Fedora but if it won't install, then there is no path to that choice.
Maybe the work for the Raspberry Pi could be looked at as an option for smaller machines.
But the direction Fedora isn't to look backwards but forwards.
On 2014-03-26 03:12, Suvayu Ali wrote:
On Wed, Mar 26, 2014 at 12:14:44AM -0600, Robin Laing wrote:
I would really like to see Fedora move to a constant flux where applications are developed and then pushed out as updates which are actually upgrades.
It is Fedora policy to not do that:
I know this and I don't like it when there is a broken package in use on Fedora with submitted bug reports related to the broken features. This is why I would like to see it changed.
It is hard to sell someone on a product when it is broken and will remain broken for months do to policy.
Thanks for the link, I hadn't seen this one before.
Get to a point of not having a F19 or F20 but just Fedora. You move from 19 to 20 to 21 through the daily updates instead of a major upgrade.
You are talking about a rolling release, this topic has been discussed to death and rejected.
When I had decent access to this list at work, I did follow the discussion. I would still like to see it changed but that is my thoughts. It would save me time and headaches.
Cheers,
Now for a pint.
Hi
On Fri, Mar 28, 2014 at 12:09 AM, Robin Laing wrote:
On 2014-03-26 03:12, Suvayu Ali wrote:
It is Fedora policy to not do that:
I know this and I don't like it when there is a broken package in use on Fedora with submitted bug reports related to the broken features. This is why I would like to see it changed.
That really depends on the importance of the broken features vs the risk of changes introducing new problems. You should recognize that trade-off. If a fix can be provided in a targeted way, there is no need to change any policy here. It cannot be a free for all type of deal
Rahul
Tim:
When I was young, we didn't have no computers. We used pencil and paper, and we were grateful. ;-)
Rick Stevens:
And you probably ate dirt and carpet fuzz as well.
Only when we were lucky.
Chris Murphy wrote:
On Mar 25, 2014, at 2:41 AM, lee lee@yun.yagibdah.de wrote:
Chris Murphy lists@colorremedies.com writes:
What I ought to do is not QA Manual Partitioning anymore and just let the people who actually think they need or want it do all the testing and bug reporting for it and let it become whatever it becomes.
Most of the size reporting problems like this are non-contiguous sections of free space being added up and reported as Available space; but the request is for a partition size greater than the largest contiguously available space.
Maybe it begins with the installer messing together all the disks in some weird way rather than to treat them separately and just let you partition them the way you want to. IIRC, there wasn`t even a way to tell it which partition to put where.
It is possible although the UI isn't obvious. You click on the mount point in question, and then there is this 3rd or 4th button under the mount points section that looks like a wrench and screwdriver (?) that you click. And that brings up a dialog where you choose which drive that mount point's underlying partition appears on.
You say this as if having to dig down to get at what should be the first thing was reasonable. Even you admit it isn't obvious. Why not? It was.
The makers of the installer can always look into this list and see what ppl say about the installer and learn from that.
No they will not do this, and it's inappropriate to even suggest it. That you don't get that simply means you're ignorant of how the process works.
The process does not include listening to users.
Bug reports are not suited for this, and complaining that ppl don`t make enough of them doesn`t get you anywhere.
Filing a bug report is the process. That's it. It works this way for everything: gnome, kde, and even commercial projects do it this way. They do not have developers hanging out in user forums ever. Sometimes QA people hang out in user forums.
So how will filing a bug report help, when the issue isn't that it fails to perform as intended, but that the intended UI is, by design,in-obvious and hard to use?
This is how I know your problem isn't really serious, because it's so unimportant to you, you won't lift a finger to contribute to any improvement. Why should I help you since you won't even help yourself?
How much attention and fixing do you think that would get? It`ll probably be closed with WONTFIX, one reason being that what is said here doesn`t refers much more to installers from F17 to F19.
It will be closed with NOTABUG, because it works as intended. Poor design is not a bug.
Those two above sentences are as close as memory allows to a direct quote of what a developer told me on another matter.
Perhaps the installer of F20 has be re-designed from scratch, at least when it comes to partitioning, and now works fine.
The installer is redesigned as of Fedora 18.
And therefore a bug report suggesting a rededign (or roll back to the old, easy to use,) installer is not happening.
you just do country and keyboard setup --- which is missing in Fedoras installer, there was no way to tell it that I have a German keyboard ---
Fedora 18, 19 and 20 have a keyboard spoke in the installer which is how you tell it you want to use a German keyboard layout.
Whatever they might have, I tried several times (because I had to start over many times because it refused to do the partitioning) to tell the installer of F19 that I have a German keyboard, and there was no way.
OK well I just tried it on Fedora 19 Netinstall and Fedora 20 Live Desktop and I could easily choose a German keyboard, so I don't know what you're doing wrong. It's right there under Localization in the Keyboard option on the main menu right after choosing language.
I don`t even know what you mean by "spoke". You boot the life system, search for "install" to find the installer, then you get an icon and start the installer, and pretty soon you get stuck with trying to do the partitioning.
Hub = the main menu that comes after language selection. On the hub are "spokes" which are paths for various things like "Time & Date" and "Keyboard" and "Installation Destination".
In English we would call that a "menu item"
Powell, Michael wrote:
It doesn`t give you choices. It leaves you in the dark about that it is somehow possible to use an non-gui installer and to do a minimal install. It leaves you in the dark about what exactly happens when you do the partitioning and with trying to figure out how get the partitioning you want.
As I said before, the entire installer was rebuilt to promote a more laid back approach. The user can choose the order they wish to customize / experience the GUI installation instead of being forced down a specific path. There might be a couple mandatories, but for the most part it's all about choice.
A non-gui installation is not something that the majority of users will choose so it's not apparent, but if you want that method, here you go: http://docs.fedoraproject.org/en-US/Fedora/20/html/Installation_Guide/ch-gui...
Thank you, I'll look at this. It's also possible some of the issues mentioned come from installing off a live CD instead of an install disk. I'm assuming that what "life disk" meant...
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
I definitely think the usability of the partitioning scheme in the current installer needs work, but as I stated earlier, I think some people are just griping about the change instead of there actually being an issue; although, it does sounds like you actually had an issue with it not detecting the correct size / free space of your drives. You might want to submit a defect, because in comparison, I have a multi-drive setup with LUKs encryption and I can have both drives wiped and start all over with encryption in less than a minute or if I want to retain one disk scheme but clobber the other it takes about 3 minutes.
On this we agree, it needs a manual partition mode which really does just that, allows the user to create and remove partitions, share partitions between release versions installed, etc. The "old installer" wasn't perfect at this, I have booted live {CD,USB} and partitioned the drives manually, as mentioned above, but that was some really complex layout. I don't consider sharing /home and SWAP between two bootable releases to be so complex it must be done in rc.local.
The computer industry knows that more and more people want installations to be less scary and faster. This trend has been seen in the evolution of the Windows installer, MacOS, most Linux distros, and even iOS or Android. There are going to be some Linux distros that don't embrace this, you mentioned Debian and I'm sure slackware and Arch won't either, but in the end it's all about attracting more users to the product. The choices are there, but us hardcore users just have to look more since we're the minority.
One of my issues with Fedora is that at times the "make it easier" flag is waved, and other times the marching song is "we are cutting edge." There really seems to be division in the policy, or an assumption that "choice" and "easy" are somehow mutually exclusive.
Easier install means a good default, and clear unambiguous ways to do complex configurations. Note, I don't suggest that "simple" or "easy" are requirements, but ways to see the raw hardware clearly and tell the OS how to utilize it seem to fit "unambiguous" perfectly.
Joe Zeff wrote:
On 03/25/2014 12:10 PM, Powell, Michael wrote:
I disagree; if a user is presented with the following filesystem choices, btrfs, ext2, ext3, ext4, JFS, reiser4, reiserFS, and ZFS, and each is presented equally with a single paragraph describing its benefits, unless the user has prior knowledge about what is the best choice for the intended installation goal, they're most likely going to spend a great deal of time reading each paragraph. If you simplify the choices to 4 instead of 8, the user has less paragraphs to read and can make a decision faster.
Agreed. However, it might be a good compromise to list only the most common selections, with a button marked Advanced Options (There's probably a better label, but if so, I can't think of it right now.) containing the rest. Beginners will, most likely, avoid that, but those who need/want the more specialized file systems will still have access to them.
This is absolutely just what is needed. Any hope it could get into fc21? If I could pick three things for some future release, they would be: - better install for controlling allocation of storage by device (but keep the serial number visible!!) - install without using network, even if available (because slow nets and/or per-byte charges prohibit upgrade to current) - a more minimal "minimal install" would be nice.
Anyway, having access to full control, somewhere easy to find, would probably solve many complaints. Thanks for listening.
On 03/28/2014 09:47 AM, Bill Davidsen issued this missive:
Joe Zeff wrote:
On 03/25/2014 12:10 PM, Powell, Michael wrote:
I disagree; if a user is presented with the following filesystem choices, btrfs, ext2, ext3, ext4, JFS, reiser4, reiserFS, and ZFS, and each is presented equally with a single paragraph describing its benefits, unless the user has prior knowledge about what is the best choice for the intended installation goal, they're most likely going to spend a great deal of time reading each paragraph. If you simplify the choices to 4 instead of 8, the user has less paragraphs to read and can make a decision faster.
Agreed. However, it might be a good compromise to list only the most common selections, with a button marked Advanced Options (There's probably a better label, but if so, I can't think of it right now.) containing the rest. Beginners will, most likely, avoid that, but those who need/want the more specialized file systems will still have access to them.
This is absolutely just what is needed. Any hope it could get into fc21? If I could pick three things for some future release, they would be:
- better install for controlling allocation of storage by device (but keep the serial number visible!!)
- install without using network, even if available (because slow nets and/or per-byte charges prohibit upgrade to
current)
- a more minimal "minimal install" would be nice.
Anyway, having access to full control, somewhere easy to find, would probably solve many complaints. Thanks for listening.
This really needs to be aimed at the fedora-test list, since that's where the development stuff is going on. ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks@alldigital.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - Sarchasm: The gulf between the author of sarcastic wit and the - - reader...who doesn't get it. - ----------------------------------------------------------------------
Paul Cartwright wrote:
On 03/26/2014 01:32 AM, Robin Laing wrote:
I`m not an expert with Fedoras installers in any way. This is simply my "user experience". Maybe the "user experience" the installer provides should be different.
It is hidden. I do use it.
If you were trying the live disk, then I believe that the manual partitioning isn't there. I was told on this list to use the full DVD. Yes, I use DVD as an install medium. There is a reason.
I came from Debian recently with Fedora 20. I was told to use the net install ISO, I did. I have windows and another linux distro installed and used manual partitioning without a problem. I have 2 drives and used my existing /home on drive A and installed Fedora on drive B. works fine.
I can only say that I am in awe of your ability to do this using the installer. After many attempts to get this working, I have taken to install with a dummy user, then edit fstab to get the /home mount, then mounting, then adding the user(s) with the UID they need to have and using the existing home directory for each.
Tim wrote:
Allegedly, on or about 21 March 2014, Powell, Michael sent:
The system name or hostname is important to networking; so, I can see why it's under networking, but I believe your frustration is more related to the lack of guidance and quality than anything else.
Well, actually, for a lot of people, the system name is simply what they want to call the computer. The computer may not even be on a network, at all. There's certainly cause for having a process of naming the computer.
For LANs, the hostname may not actually be set on the computer. A DHCP server may dole out IPs and hostnames, or simply dole out hostnames and a DNS lookup discovers what name is given to that IP. In that scenario there's some logic to using the network configuration to fill in the name. Though there's still the reverse case, where I want to name a computer, and let the rest of the network find out what my name is, from me (well, /this/ computer, not actually me, personally).
This is useful in a large site, but also in virtual machines. I have a single VM image which works thus: - set MAC on startup - DHCP returns IP and name from MAC - rc.local uses the name to start services on the VM, dhcp, mail, nntp, dns, pop, etc. - can include dedicated user machine or app, network mounting other partitions as needed.
Chris Murphy wrote:
On Mar 21, 2014, at 6:18 PM, Liam Proven lproven@gmail.com wrote:
Secondly, I can confirm this finding. I was completely unable to install F20 using the current installer program. My system has 2 drives - a 1TB HD and a 120GB SSD. The SSD holds Ubuntu and Win7; the 1TB drive holds /home, swap, a dedicated Windows swap partition, a Windows data drive, and 2 spare unused root partitions for test distros.
This is /not/ a very complex layout - there is no RAID, no LVM, no GPT, nothing hairy or difficult.
It is actually a complex layout. Most of the world's installers can't deal with what you just described.
I will simply note that the old installer could...
[...discussion of unrelated OS behavior...]
And yes, it's fair to bring up what money bags companies with more money and resources than god. Because this is an area where they've all considered what you're describing, is an edge case. Not common at all.
Really? On a percentage basic, I assume you're right. But in terms of raw number of the current Fedora user base, I would guess that 30% of the users have another OS installed, and almost as many have multiple drives. Getting a 120GB SSD for <$100 these days is causing many people to rethink. So the number of people who do or may see this is probably in the thousands. It certainly isn't rare.
The Fedora 20 installer's default/easy/guided/auto path installs to free space. Yet it has more options and outcomes than the total number of all possible options in both the Windows and OS X installers combined.
That's definitely true, my personal experience indicates that many of the outcomes are suboptimal (meaning fail to boot ANYTHING after install, delete of otherwise lose data on existing partitions, or won't boot anything except Fedora).
Hmm. Now I believe you were just about to cite a bugzilla ID describing the above behavior?
It's really impractical to list all the failure modes, and probably not helpful id someone did. Would a description of seemingly valid actions resulting in failure repeated in N variations help or swamp the system, getting no improvement?
Would a suggestion be valid in a bug report, which said: if the user specifies a partition (meaning valid storage unit) with a known filesystem installed the installer should default to "use, don't lose" and present [use] [reformat] [oops] options before proceeding.
The F20 installer was completely unable to understand it and allow me to install a complete system. Assigned some 250GiB of space, it said that it needed 6.5GB and there wasn't enough room.
I've done hundreds of hours of installer testing over the last year. It has been really frustrating. This is the most complicated/capable installer I've ever worked with other than maybe the OpenSUSE installer. Out of the gate it offerred too much compared to the time/resources allotted for QA, debugging, and code changes needed.
The reality is, you get either stability or you get features. You don't get both. The mantra for the new installer was about getting as many of old installer's features into the new one as soon as possible, and stability was simply expected to have to take a hit in order to do that. And that's exactly what happened.
It seems that there are a number of users, some of us experienced users, who feel that the current "new" installer is lacking in both. It's hard to find a way to tell it to do what you want, and when you do in many cases it doesn't work properly.
Let's pretend the installer could only do 20% of what the old installer did, yet it was almost bullet proof - never crashed, didn't have any of the logic problems you're talking about, and so on. Would Fedora users have understood that trade off? Maybe a lot of them would have. But then we'd have a lot of others pining for a right to a GUI that lets them create some of the most esoteric storage layouts of all time.
The first thing users don't understand is why things in Fedora are rewritten from scratch instead of improved. Evolution rather than replacement. Fedora dinosaurs don't improve until they become birds, they are whacked with a meteor and replaced with mammal 0.9, a not ready for existence in a harsh world, premature baby. And when that becomes reliable and usable, WHACK! comes the next meteor.
And guess what? That has to be coded, and ostensibly should be tested. And quite frankly the QA resources are really limited. Not every possible combination permitted in Manual Partitioning is tested at all. That's how much it can do. It's nearly unlimited possibilities because, guess another thing, I've never once seen it disqualify a drive layout from the start. I've never seen it look at a crazy layout and go "umm yeah, no please use gparted and obliterate this drive first." But I've seen that many times with the OS X installer: flat out refusal, "go format the drive in Disk Utility." Quite a few times when trying to prepare a drive for dual boot on OS X I've seen the error message that the disk can't be partitioned, and that I had to obliterate the whole drive and reinstall OS X from scratch in order to install Windows side by side. So really, anaconda is extremely tolerant and I think that's something of a problem too. It probably should be disqualifying a lot of nut case layouts, and just saying no.
And would it be easier to test improvements on an existing project (yes), or more responsible not to release software which doesn't work (yes, again)?
In trying to install, it erased one of the spare-root partitions and was unable to recreate it in the available empty space.
And you have a bug for this? It's *really* difficult to get the installer to inadvertently delete partitions. It requires two clicks: selection, then deletion. For guided partitioning, the button is labeled "delete" whereas the button in manual partitioning is labeled as a minus symbol.
Is there a case where a partition of a certain partition type will be considered "up for grabs" and used without asking? Real question, not being snarky, I think I've seen this as well, the boot partition of one install was reused as the boot for a newer total install, and had it asked I would not have clicked a [reuse-partition] button.
One thing that some people don't easily grok is that Manual Partitioning isn't partitioning oriented.
Why not? And why isn't there a mode which is, so the user can set up ther system as they need it?
It's mount point oriented. And that's because
mount points can be partitions, subvolumes, logical volumes, or md block devices. It's not correct to call all of those things partitions. But all of them ultimately are assigned mount points. This is a top down view, rather than the typical bottom up view where you always have partitions, and then maybe you have raid devices or LVs or subvolumes. The idea is to think less about the details of the layout and more about the outcome you want.
It doesn't seem to know about using nbd, either, or creating array with a deliberately missing member (nbd to be added later, write mostly). I don't fault the lack of nbd support, I do the missing way to create RAID with missing members.
Did the install team ever consider just letting people drop into a shell and use command line tools to do a tiny bit of magic, then rescan the physical devices again to be sure the configuration is understood? It might save a huge pile of user frustration and cries for support of flexible configuration options.
This is understandably confusing if you're really familiar with storage stack creation. But most people aren't. Nevertheless, one of the first steps is drive selection, which is about the most bottom layer there is. And then the very next step is the top most layer, which are the mount points. So it's an unexpected context shift from bottom to top, seemingly without any conversation about what's happening in the middle. It's a different approach.
If you remain attached that what you're doing in Manual Partitioning is in fact partitioning, you'll continue to be frustrated.
Then why not provide manual partitioning, or rename the mode to something "pretending you have any say in how we do install?" Frustrated is exactly the issue here.
Now, if you didn't file a bug about your anecdote, I want you to imagine me staring at you with a look of "really?" Because this much effort complaining yet no bug report? How exactly do you expect it to get better?
I've been a user since fc3 or 4, and I have learned that complaints about design issues are treated as suggestions for future enhancement rather than bugs. If the software functions as designed it's not a bug, it's a design decision. I have said that on (rare) occasions, although I try to say "the software doesn't currently do that."
Tom Horsley wrote:
On Sat, 22 Mar 2014 16:40:36 +0000 Liam Proven wrote:
Meantime, for further Fedora eval, it's going in a VirtualBox. Sad, but that's all it seems able to handle.
Actually, that's the key to installing on disk in a sensibly created partition layout: Install in a virtual machine 1st, then copy the VM disk image to a real disk and fiddle with the grub config and fstab. I install that way all the time now since the new installer appeared, and find it actually make things lots more convenient since I don't have to have the machine down while I'm doing the install.
Because of the write speed of USB sticks, I find that's a good way to do a portable install as well, create a drive image and install on that. When done, then you just copy to the USB stick.
On 03/25/2014 02:24 PM, Chris Murphy wrote:
On Mar 25, 2014, at 2:41 AM, lee lee@yun.yagibdah.de wrote:
Chris Murphy lists@colorremedies.com writes:
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
Please post the bugzilla URL.
I didn`t make one, I was busy trying to install. You need to get used to that not every problem is or can be made into a bug report, especially not one that would be in any way useful.
What I ought to do is not QA Manual Partitioning anymore and just let the people who actually think they need or want it do all the testing and bug reporting for it and let it become whatever it becomes.
Most of the size reporting problems like this are non-contiguous sections of free space being added up and reported as Available space; but the request is for a partition size greater than the largest contiguously available space.
Maybe it begins with the installer messing together all the disks in some weird way rather than to treat them separately and just let you partition them the way you want to. IIRC, there wasn`t even a way to tell it which partition to put where.
It is possible although the UI isn't obvious. You click on the mount point in question, and then there is this 3rd or 4th button under the mount points section that looks like a wrench and screwdriver (?) that you click. And that brings up a dialog where you choose which drive that mount point's underlying partition appears on.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
It isn't going to get better complaining about it on this list. Do you have bugzilla IDs, and if so post them. If not, then how do you expect the behavior to get any better? Magic?
The makers of the installer can always look into this list and see what ppl say about the installer and learn from that.
No they will not do this, and it's inappropriate to even suggest it. That you don't get that simply means you're ignorant of how the process works.
Bug reports are not suited for this, and complaining that ppl don`t make enough of them doesn`t get you anywhere.
Filing a bug report is the process. That's it. It works this way for everything: gnome, kde, and even commercial projects do it this way. They do not have developers hanging out in user forums ever. Sometimes QA people hang out in user forums.
I believe what you want is a request for enhancement(RFE) because it probably works as intended. A clear succinct explanation of what the perceived problem is and what the suggested enhancement is or should be has to included. Saying you don't like it isn't going to get you anywhere. You have to lucidly explain why it falls short and how to improve it. You have to appeal to logic not emotion.
Or, since you keep insisting on bug reports, why don`t you go ahead and put together a list of URLs to the list archive pointing to posts about the installer to compile a combined bug report?
Why don't you go ahead and send me 4-5 bitcoins and I'll think about it?
This is how I know your problem isn't really serious, because it's so unimportant to you, you won't lift a finger to contribute to any improvement. Why should I help you since you won't even help yourself?
How much attention and fixing do you think that would get? It`ll probably be closed with WONTFIX, one reason being that what is said here doesn`t refers much more to installers from F17 to F19.
Perhaps the installer of F20 has be re-designed from scratch, at least when it comes to partitioning, and now works fine.
The installer is redesigned as of Fedora 18.
you just do country and keyboard setup --- which is missing in Fedoras installer, there was no way to tell it that I have a German keyboard ---
Fedora 18, 19 and 20 have a keyboard spoke in the installer which is how you tell it you want to use a German keyboard layout.
Whatever they might have, I tried several times (because I had to start over many times because it refused to do the partitioning) to tell the installer of F19 that I have a German keyboard, and there was no way.
OK well I just tried it on Fedora 19 Netinstall and Fedora 20 Live Desktop and I could easily choose a German keyboard, so I don't know what you're doing wrong. It's right there under Localization in the Keyboard option on the main menu right after choosing language.
I don`t even know what you mean by "spoke". You boot the life system, search for "install" to find the installer, then you get an icon and start the installer, and pretty soon you get stuck with trying to do the partitioning.
Hub = the main menu that comes after language selection. On the hub are "spokes" which are paths for various things like "Time & Date" and "Keyboard" and "Installation Destination".
Chris Murphy
On 26.03.2014 23:58, Patrick O'Callaghan wrote:
On Wed, 2014-03-26 at 22:42 +0000, Timothy Murphy wrote:
David G. Miller wrote:
The funny thing is that back in the earliest days of Unix, /usr is where user directories lived. When K&R ran out of room in / for programs, they looked to for a partition that had additional space available and it was /usr. Originally programs ended up in /usr/bin simply because there wasn't room for them in /bin; not for some usage reason.
Yes, I recall that the first Unix system I ran, version 5 on a pdp-11/23, had two (enormous) 10MB disks, one for the kernel and the other /usr .
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
poc
No matter how small disk can be, can't beat a punched card.
poma
On Sat, 2014-03-29 at 03:59 +0100, poma wrote:
On 26.03.2014 23:58, Patrick O'Callaghan wrote:
On Wed, 2014-03-26 at 22:42 +0000, Timothy Murphy wrote:
David G. Miller wrote:
The funny thing is that back in the earliest days of Unix, /usr is where user directories lived. When K&R ran out of room in / for programs, they looked to for a partition that had additional space available and it was /usr. Originally programs ended up in /usr/bin simply because there wasn't room for them in /bin; not for some usage reason.
Yes, I recall that the first Unix system I ran, version 5 on a pdp-11/23, had two (enormous) 10MB disks, one for the kernel and the other /usr .
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
poc
No matter how small disk can be, can't beat a punched card.
Been there, done that. Also patched paper-tape to fix bugs.
poc
On 29.03.2014 12:48, Patrick O'Callaghan wrote:
On Sat, 2014-03-29 at 03:59 +0100, poma wrote:
On 26.03.2014 23:58, Patrick O'Callaghan wrote:
On Wed, 2014-03-26 at 22:42 +0000, Timothy Murphy wrote:
David G. Miller wrote:
The funny thing is that back in the earliest days of Unix, /usr is where user directories lived. When K&R ran out of room in / for programs, they looked to for a partition that had additional space available and it was /usr. Originally programs ended up in /usr/bin simply because there wasn't room for them in /bin; not for some usage reason.
Yes, I recall that the first Unix system I ran, version 5 on a pdp-11/23, had two (enormous) 10MB disks, one for the kernel and the other /usr .
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
poc
No matter how small disk can be, can't beat a punched card.
Been there, done that. Also patched paper-tape to fix bugs.
poc
Whence the 'patch' comes. :) Pat, you are a unixsaur. :)
poma
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
Ha! At FAU the engineering department had a Data General Nova with only 8KB of core memory, and an ASR-33 TTY with a paper tape attachment as the only I/O device :-).
On Sat, 2014-03-29 at 11:01 -0400, Tom Horsley wrote:
That ain't nuttin'. We started with an 11/45 with 48KB of core memory and 2 RK05's (2.2MB removable cartridge disks).
Ha! At FAU the engineering department had a Data General Nova with only 8KB of core memory, and an ASR-33 TTY with a paper tape attachment as the only I/O device :-).
Had one of those in my first job. The OS was so terrible the guy I worked for wrote his own multitasking kernel for it, which we used for one of the first computer typesetting systems (Cambridge University Press 1972 :-) All done in BCPL (ancestor of C).
poc
On 2014-03-28 03:20, Rahul Sundaram wrote:
Hi
On Fri, Mar 28, 2014 at 12:09 AM, Robin Laing wrote:
On 2014-03-26 03:12, Suvayu Ali wrote:
It is Fedora policy to not do that:
<http://fedoraproject.org/wiki/Updates_Policy#Philosophy>
I know this and I don't like it when there is a broken package in use on Fedora with submitted bug reports related to the broken features. This is why I would like to see it changed.
That really depends on the importance of the broken features vs the risk of changes introducing new problems. You should recognize that trade-off. If a fix can be provided in a targeted way, there is no need to change any policy here. It cannot be a free for all type of deal
Rahul
In the long run, I agree with you on this matter.
When a feature that is used is broken, even to the point an application crashes, and the fix requires the new version, then it is in issue in the usability of Fedora. This has been my biggest complaint over the years. I am finding this with some applications on F20 that I have used in the past. In some cases, it isn't even a full version update but a 0.1 update that is refused.
From an application issue, it shouldn't affect other programs or applications at all as the build is based on the libraries that are already there.
I can agree on changing a library or other main base for the rest of Fedora as the issues can be a major problem.
Program like pam_mount when security could be compromised are also important to see fixed. I ran into this some years ago with encrypted home/[user] directories. I couldn't get any support upstream because the Fedora version used was to old, even though the current version or Fedora was the latest.
In the past, in order to get a working application, I have even downloaded the source code and compiled it to continue working. I need to learn how to rebuild rpm's but I never seem to have the time.
You and I are not going to change this.
Robin
On 2014-03-28 10:54, Bill Davidsen wrote:
Paul Cartwright wrote:
On 03/26/2014 01:32 AM, Robin Laing wrote:
I`m not an expert with Fedoras installers in any way. This is simply my "user experience". Maybe the "user experience" the installer provides should be different.
It is hidden. I do use it.
If you were trying the live disk, then I believe that the manual partitioning isn't there. I was told on this list to use the full DVD. Yes, I use DVD as an install medium. There is a reason.
I came from Debian recently with Fedora 20. I was told to use the net install ISO, I did. I have windows and another linux distro installed and used manual partitioning without a problem. I have 2 drives and used my existing /home on drive A and installed Fedora on drive B. works fine.
I can only say that I am in awe of your ability to do this using the installer. After many attempts to get this working, I have taken to install with a dummy user, then edit fstab to get the /home mount, then mounting, then adding the user(s) with the UID they need to have and using the existing home directory for each.
Practice, mistakes and crossed fingers. :)
It can be done but you have to be careful about it. I find that it is easier to install a base system and then modify the mount points as needed. As I use pam_mount for [users] home directories, this makes some aspects easier. That is a post install configuration anyways.
I use a basic /home on the root drive with each users directory mounted to their username mount point.
On 2014-03-28 14:23, Max wrote:
On 03/25/2014 02:24 PM, Chris Murphy wrote:
On Mar 25, 2014, at 2:41 AM, lee lee@yun.yagibdah.de wrote:
Chris Murphy lists@colorremedies.com writes:
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
Please post the bugzilla URL.
I didn`t make one, I was busy trying to install. You need to get used to that not every problem is or can be made into a bug report, especially not one that would be in any way useful.
What I ought to do is not QA Manual Partitioning anymore and just let the people who actually think they need or want it do all the testing and bug reporting for it and let it become whatever it becomes.
Most of the size reporting problems like this are non-contiguous sections of free space being added up and reported as Available space; but the request is for a partition size greater than the largest contiguously available space.
Maybe it begins with the installer messing together all the disks in some weird way rather than to treat them separately and just let you partition them the way you want to. IIRC, there wasn`t even a way to tell it which partition to put where.
It is possible although the UI isn't obvious. You click on the mount point in question, and then there is this 3rd or 4th button under the mount points section that looks like a wrench and screwdriver (?) that you click. And that brings up a dialog where you choose which drive that mount point's underlying partition appears on.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
It isn't going to get better complaining about it on this list. Do you have bugzilla IDs, and if so post them. If not, then how do you expect the behavior to get any better? Magic?
The makers of the installer can always look into this list and see what ppl say about the installer and learn from that.
No they will not do this, and it's inappropriate to even suggest it. That you don't get that simply means you're ignorant of how the process works.
Bug reports are not suited for this, and complaining that ppl don`t make enough of them doesn`t get you anywhere.
Filing a bug report is the process. That's it. It works this way for everything: gnome, kde, and even commercial projects do it this way. They do not have developers hanging out in user forums ever. Sometimes QA people hang out in user forums.
I believe what you want is a request for enhancement(RFE) because it probably works as intended. A clear succinct explanation of what the perceived problem is and what the suggested enhancement is or should be has to included. Saying you don't like it isn't going to get you anywhere. You have to lucidly explain why it falls short and how to improve it. You have to appeal to logic not emotion.
I agree. Some are easy as adding a word here or there. On the Installation Destination screen, when you select the drive, it not say "Done" on the button but "Next" or "Continue" as there is another screen. This is confusing the first time you run into it.
On the next screen. adding the word customize after review/modify, will also help clarify the choice for customized installation. Modify doesn't mean customize to me.
RFE's on bugzilla. 1079655 and 1082890 cover these two points.
One thing I have learned over the years in most of my work. When writing instructions, write them for someone two grade levels below you and you should cover all aspects of the task at hand.
Robin
Hi
On Mon, Mar 31, 2014 at 11:24 PM, Robin Laing wrote:
When a feature that is used is broken, even to the point an application crashes, and the fix requires the new version, then it is in issue in the usability of Fedora. This has been my biggest complaint over the years. I am finding this with some applications on F20 that I have used in the past. In some cases, it isn't even a full version update but a 0.1 update that is refused.
Did you file any bug report? Example? In any case, changing the update policy is not going to fix this. There is nothing in the update policy that prevents a maintainer from pushing a point release that fixes a bug. It could be resource shortage issue but a policy will never force volunteer maintainers. It can only set guidelines that communicate expectations.
Rahul
On 03/31/2014 09:01 PM, Robin Laing wrote:
I use a basic /home on the root drive with each users directory mounted to their username mount point.
So each user has their own partition? Do you also use LVM so that you can create/resize or remove them as needed? That's an interesting idea.
On Mon, 2014-03-31 at 22:01 -0600, Robin Laing wrote:
On 2014-03-28 10:54, Bill Davidsen wrote:
Paul Cartwright wrote:
On 03/26/2014 01:32 AM, Robin Laing wrote:
I`m not an expert with Fedoras installers in any way. This is simply my "user experience". Maybe the "user experience" the installer provides should be different.
It is hidden. I do use it.
If you were trying the live disk, then I believe that the manual partitioning isn't there. I was told on this list to use the full DVD. Yes, I use DVD as an install medium. There is a reason.
I came from Debian recently with Fedora 20. I was told to use the net install ISO, I did. I have windows and another linux distro installed and used manual partitioning without a problem. I have 2 drives and used my existing /home on drive A and installed Fedora on drive B. works fine.
I can only say that I am in awe of your ability to do this using the installer. After many attempts to get this working, I have taken to install with a dummy user, then edit fstab to get the /home mount, then mounting, then adding the user(s) with the UID they need to have and using the existing home directory for each.
Practice, mistakes and crossed fingers. :)
It can be done but you have to be careful about it. I find that it is easier to install a base system and then modify the mount points as needed. As I use pam_mount for [users] home directories, this makes some aspects easier. That is a post install configuration anyways.
I use a basic /home on the root drive with each users directory mounted to their username mount point.
I agree as a first time Fedora user, coming from Kubuntu I found the graphical installer of F20 a bit puzzling. because coming in it saw that I had Kubuntu already installed and had a big ! over the partitioning and I had to click on it then click on my hard disk then click the back button at the upper left corner to go back to the main installation screen.
first time it didnt take my selection to erace the whole hdd... I had to go back and click on partitions double click the hdd read though the message saying not enough space then I clicked blindly the use basic /home setup. and hoped for the best :X
Christopher
On 04/01/2014 12:10 AM, Robin Laing wrote:
On 2014-03-28 14:23, Max wrote:
On 03/25/2014 02:24 PM, Chris Murphy wrote:
On Mar 25, 2014, at 2:41 AM, lee lee@yun.yagibdah.de wrote:
Chris Murphy lists@colorremedies.com writes:
Partitioning took me about three hours with the installer of F19, with a very simple setup and not even data to preserve and neither RAID, nor encryption, and it was only possible after I created the partitions outside the installer. There was no way to do it with the installer, it kept saying there isn`t enough room despite there was plenty, and it did what it wanted rather than what I wanted.
Please post the bugzilla URL.
I didn`t make one, I was busy trying to install. You need to get used to that not every problem is or can be made into a bug report, especially not one that would be in any way useful.
What I ought to do is not QA Manual Partitioning anymore and just let the people who actually think they need or want it do all the testing and bug reporting for it and let it become whatever it becomes.
Most of the size reporting problems like this are non-contiguous sections of free space being added up and reported as Available space; but the request is for a partition size greater than the largest contiguously available space.
Maybe it begins with the installer messing together all the disks in some weird way rather than to treat them separately and just let you partition them the way you want to. IIRC, there wasn`t even a way to tell it which partition to put where.
It is possible although the UI isn't obvious. You click on the mount point in question, and then there is this 3rd or 4th button under the mount points section that looks like a wrench and screwdriver (?) that you click. And that brings up a dialog where you choose which drive that mount point's underlying partition appears on.
It was seriously awful. It would have taken 10--15 minutes with the Debian installer.
It isn't going to get better complaining about it on this list. Do you have bugzilla IDs, and if so post them. If not, then how do you expect the behavior to get any better? Magic?
The makers of the installer can always look into this list and see what ppl say about the installer and learn from that.
No they will not do this, and it's inappropriate to even suggest it. That you don't get that simply means you're ignorant of how the process works.
Bug reports are not suited for this, and complaining that ppl don`t make enough of them doesn`t get you anywhere.
Filing a bug report is the process. That's it. It works this way for everything: gnome, kde, and even commercial projects do it this way. They do not have developers hanging out in user forums ever. Sometimes QA people hang out in user forums.
I believe what you want is a request for enhancement(RFE) because it probably works as intended. A clear succinct explanation of what the perceived problem is and what the suggested enhancement is or should be has to included. Saying you don't like it isn't going to get you anywhere. You have to lucidly explain why it falls short and how to improve it. You have to appeal to logic not emotion.
I agree. Some are easy as adding a word here or there. On the Installation Destination screen, when you select the drive, it not say "Done" on the button but "Next" or "Continue" as there is another screen. This is confusing the first time you run into it.
On the next screen. adding the word customize after review/modify, will also help clarify the choice for customized installation. Modify doesn't mean customize to me.
RFE's on bugzilla. 1079655 and 1082890 cover these two points.
One thing I have learned over the years in most of my work. When writing instructions, write them for someone two grade levels below you and you should cover all aspects of the task at hand.
Too often we express thoughts that are born solely out of our own experience. We fail to provide all the background and context for the statement we are making and so other people do not necessarily get it if they haven't had the same experience.
So if your using custom partitioning then it may be worthwhile to explain why you use custom partitioning. Not because its anybody's business but simply because it provides context to the person to whom your trying to explain exactly how the issue arose, you want to put them in your shoes, make them forget who they are for a moment so they can see the situation through your eyes. This can provide insight into the exact nature of the issue.
Though depending on who your talking too it just might confuse them. Just think how boring the world would be if we were all on the same level!
A lot of the time we feel compelled to /argue/ our point and it becomes a competition, a power struggle of sorts. What we should be doing is explaining instead of trying to win an argument. Though how people respond often sets us down the path of argument before we realize what has happened.
If you want to convince someone of your viewpoint then you usually have to be very deliberate. Success is never guaranteed regardless of what explanations we may provide. Try to avoid conscious manipulation if at all possible.
On Sat, Mar 22, 2014 at 04:40:36PM +0000, Liam Proven wrote:
As I have said previously, I have /never/ successfully installed Fedora on actual hardware since v1.0 shipped in, what was it, 2003? I have installed Haiku, Aros, FreeBSD, PC BSD, dozens of Linux distros, Windows 2 through 8, SCO Xenix, SCO Unix, OpenSolaris, OpenVMS, FreeDOS, DR-DOS, MS-DOS, PC-DOS, OS/2 1 through eComStation 2, MacOS 6 through OS X 10.9. I am *not* a newbie and I am *not* an inexperienced inexpert fumbler.
That is toooo broad a statement. The serious problem with the UI only began with F19/20 but which we users expect will get sorted out in due time ( I hope ) but there was not such a serious UI installer problem with other Fedora's. As a very old Redhat/Fedora user, I totally disagree with you to the point that your other claims also get tainted negatively.
--
On 2 April 2014 12:01, Vikram Goyal vikigoyal@gmail.com wrote:
On Sat, Mar 22, 2014 at 04:40:36PM +0000, Liam Proven wrote:
As I have said previously, I have /never/ successfully installed Fedora on actual hardware since v1.0 shipped in, what was it, 2003? I have installed Haiku, Aros, FreeBSD, PC BSD, dozens of Linux distros, Windows 2 through 8, SCO Xenix, SCO Unix, OpenSolaris, OpenVMS, FreeDOS, DR-DOS, MS-DOS, PC-DOS, OS/2 1 through eComStation 2, MacOS 6 through OS X 10.9. I am *not* a newbie and I am *not* an inexperienced inexpert fumbler.
That is toooo broad a statement. The serious problem with the UI only began with F19/20 but which we users expect will get sorted out in due time ( I hope ) but there was not such a serious UI installer problem with other Fedora's. As a very old Redhat/Fedora user, I totally disagree with you to the point that your other claims also get tainted negatively.
I too am a "very old Red Hat user" - like you, I started in the mid-1990s. I have successfully used -- and reviewed in print -- Red Hat 4, 5, 6, 7, 8 and 9, plus I've also worked with CentOS 4, 5 and 6. The first Linux distro I tried was Slackware and the first that I successfully installed was Lasermoon Linux-FT, the first LiveCD, with an innovative mechanism of caching the decompressed contents of the compressed files from the CD on the hard disk until you had a full system installed.
I remember the hoo-ha when Linux 2.0 shipped. Heck, I remember Linux 1.0 being released, although I wasn't using it back then -- in those days, I was a SCO man.
It is just Fedora I've had such consistent problems with. I haven't tried it since about F17 so I am happy to take your word that the partitioner issues are new. However, other issues prevented me getting working installs on the previous versions. As I've said before, no, I have not tried _all_ of them - I tend to come back for a fresh look every 2-3 years.
I've had problems with lots of distros! Debian used to be a horror to install; Slackware and FreeBSD still are. Mandrake once wiped all the partitions and all the other distros off a testbed PC. Gentoo compiled for 2-3 straight days and failed to build a working system, and its famed customisability did not extend to, say, letting me choose my preferred init system.
The first distro I adopted full-time was Caldera OpenLinux; later I switched to SuSE, and from SuSE to Ubuntu, which has been my main OS since 2004 except for periods of Mac use. I also try out each new version of Windows when released for about 2 months, before switching back to Unix with a sigh of relief.
On 2014-03-31 22:42, Rahul Sundaram wrote:
Hi
On Mon, Mar 31, 2014 at 11:24 PM, Robin Laing wrote:
When a feature that is used is broken, even to the point an application crashes, and the fix requires the new version, then it is in issue in the usability of Fedora. This has been my biggest complaint over the years. I am finding this with some applications on F20 that I have used in the past. In some cases, it isn't even a full version update but a 0.1 update that is refused.
Did you file any bug report? Example? In any case, changing the update policy is not going to fix this. There is nothing in the update policy that prevents a maintainer from pushing a point release that fixes a bug. It could be resource shortage issue but a policy will never force volunteer maintainers. It can only set guidelines that communicate expectations.
Rahul
There is a bug reported and the automatic reporting tool keeps telling me it is reported. I would have to look again but I think the issue is upstream and how they handle a new library.
In F19, the one application, the maintainer points to the policy of not move to a new major release.
I will live with it.
On 2014-04-01 00:00, Joe Zeff wrote:
On 03/31/2014 09:01 PM, Robin Laing wrote:
I use a basic /home on the root drive with each users directory mounted to their username mount point.
So each user has their own partition? Do you also use LVM so that you can create/resize or remove them as needed? That's an interesting idea.
No, I don't use LVM. Had a drive issue with LVM a few years ago and it was a nightmare to get around. Turned me off of LVM since. Know others that have also been burned by LVM.
With the costs of drives today, I just provide LOTS of space. If they need more, then I add more in a sub directory that they need it in.
Small system, few users, not a major management problem.
It is easy to move to a new/different partition if needed.
Using this technique and pam_mount, you can also setup, per user encrypted partitions. Extra security if needed.
I have been using this idea for over 10 years and other than an LVM headache, it has been no problem.
Robin