On Mon, Aug 11, 2014 at 05:14:52PM +0300, Elad Alfassa wrote:

The audio quality is really problematic here, especially in the second talk, and it makes it very hard to listen to the talks. Can you some up the points you want to discuss for those of us who can't watch the videos?

At some point, but in the meantime here's the slides: https://langdon.fedorapeople.org/flock-2014-preso.html

On Mon, Aug 11, 2014 at 11:56 AM, drago01 drago01@gmail.com wrote:

On Mon, Aug 11, 2014 at 5:25 PM, Matthew Miller mattdm@fedoraproject.org wrote:

...

On Mon, Aug 11, 2014 at 05:14:52PM +0300, Elad Alfassa wrote:

...

The audio quality is really problematic here, especially in the second talk, and it makes it very hard to listen to the talks. Can you some up the points you want to discuss for those of us who can't watch the videos?

At some point, but in the meantime here's the slides: https://langdon.fedorapeople.org/flock-2014-preso.html

I didn't watch the video (horrible sound) but from the slides

"Commit to holding Gnome at F21 version for 2 releases"

since when is this "current policy" ? Not only does the slides provide no justification for that it is simply a no go (holds back wayland and other efforts; user already where not really happy with the old version in F20 due to the long delays; it disconnects us from current upstream development; ...)

It isn't current policy. The slides lack the context of the talk, but they were suggestions for things we _could_ do right now. Not things we are doing right now.

josh

On Mon, 2014-08-11 at 19:08 +0300, Elad Alfassa wrote:

I also strongly oppose to not being in sync with upstream GNOME.

If we want to stay relevant, we must have an up-to-date GNOME in each new Fedora release with all the improvements and bugfixes the new GNOME releases bring.

It was just a suggestion. With some goodwill, I would interpret it as

'provide stability, both in the user experience and in the APIs that are used by application and extension developers'.

Concrete example: Make sure popular shell extensions don't break.

In other news, we've already begun to fulfill one of Langon's wishlist items:

https://wiki.gnome.org/HowDoI/SearchProvider

On Fri, Aug 15, 2014 at 5:37 AM, Adam Batkin adam@batkin.net wrote:

Unfortunately I wasn't able to watch Langdon's video, but I watched Christian's talk and read Langdon's the slides. I actually spoke with Langdon at OSCON where he accurately described me as a "developer on Linux" which is [usually] distinctly different from a "Linux developer".

To me, the biggest thing that I would like to see, is Gnome to be more comfortable for a developer (that's why we're all here, right?). Developers often spend a lot of time with their hands on the keyboard (as opposed to the mouse) and are already comfortable installing packages, etc... I'm sure Gnome is great for general computing users (though we can debate that too) and maybe touchscreen users like it, but I find it maddening. I just built a new PC though, and for the first time in a long time I'm forcing myself to use Gnome instead of XFCE (I tried KDE as recently as a few months ago and found it to still be unstable and have really weird theme/UI interactions).

GNOME works really well with a keyboard only. You can start and switch apps very easily with the keyboard "super + typepartofname + enter" ...

And no you do *not* need a touchscreen for anything. Repeating that does not make it true.

But there are a bunch of other "polish" things that I think also need to be fixed. For example:

• There are at least half a dozen Gnome Bugzillas that I've come across

around focus and window stacking issues - and I think I've encountered every one of them. This is maddening. For someone who uses the keyboard as much as possible, this makes the system almost unusable

What are those focus and stacking issues? You have to name them otherwise we can't discuss / solve them.

• The default theme makes it difficult to easily distinguish the Focused

window from all other windows - again, maddening when you have lots of windows spread across multiple very high resolution monitors. Especially with all the window focus/stacking issues

Really? At least for gtk3 apps the whole unfocsued window dims ("backdrop") which makes it very cleary shown as unfocsued. Again no idea what your "focus / stacking issues are"

• Silly things like the fact that you need to drag to get rid of the

screensaver/blanker thing - Individually, this isn't *that* important, and there's already a BZ for it, but no one seems rushing to fix it. I can't think of a single case where a non-touchscreen user would WANT that functionality, but there still doesn't appear to be a way to rid yourself of it

You do *not* need to drag it. You can just type you password. Or hit ESC or press Enter. Maybe it is not clear enough that you can do that though.

On Fri, 2014-08-15 at 12:00 +0200, Denis Washington wrote:

Am 15.08.2014 um 11:01 schrieb drago01:

...

...

• Silly things like the fact that you need to drag to get rid of the

screensaver/blanker thing - Individually, this isn't *that* important, and there's already a BZ for it, but no one seems rushing to fix it. I can't think of a single case where a non-touchscreen user would WANT that functionality, but there still doesn't appear to be a way to rid yourself of it

You do *not* need to drag it. You can just type you password. Or hit ESC or press Enter. Maybe it is not clear enough that you can do that though.

If you wake up your screen by moving the mouse, as I and probably a lot of people do, the natural thing to do is also to remove the lock screen with the mouse. That you have to drag the lock screen up to do so

Or scroll with the mouse wheel (or the edge of the touchpad, or two fingers on the touchpad)

I think this is quite a good idea. To merge the GDM login functionality into the shield itself would save both complexity and time. i.e. password field on the shield, if password is applied, else show desktop when the shield is swept away

On Fri, 15 Aug, 2014 at 1:04 PM, Adam Batkin adam@batkin.net wrote:

On 08/15/2014 06:21 AM, Mathieu Bridon wrote:

...

On Fri, 2014-08-15 at 12:00 +0200, Denis Washington wrote:

...

Am 15.08.2014 um 11:01 schrieb drago01:

...

...

• Silly things like the fact that you need to drag to get rid of

the

...

...

...

...

screensaver/blanker thing - Individually, this isn't *that*

important, and

...

...

...

...

there's already a BZ for it, but no one seems rushing to fix it.

I can't

...

...

...

...

think of a single case where a non-touchscreen user would WANT

that

...

...

...

...

functionality, but there still doesn't appear to be a way to rid

yourself of

...

...

...

...

it

You do *not* need to drag it. You can just type you password. Or

hit

...

...

...

ESC or press Enter. Maybe it is not clear enough that you can do that though.

If you wake up your screen by moving the mouse, as I and probably

a lot

...

...

of people do, the natural thing to do is also to remove the lock

screen

...

...

with the mouse. That you have to drag the lock screen up to do so

Or scroll with the mouse wheel (or the edge of the touchpad, or two fingers on the touchpad)

I don't have a password on my screensaver. I expect to nudge my mouse and see my desktop, or hit the Ctrl key and see my desktop. Instead I see the curtain and have to either drag it away or hit Esc or Enter. Again not a big deal, but annoying.

FYI: https://bugzilla.gnome.org/show_bug.cgi?id=696330

-Adam Batkin

-- desktop mailing list desktop@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/desktop

Adam Batkin píše v Pá 15. 08. 2014 v 08:19 -0400:

...

GNOME works really well with a keyboard only. You can start and switch apps very easily with the keyboard "super + typepartofname + enter" ...

Isn't that for launching a new instance of an app?

I often have 2 or 3 windows that I'm actively working with. That's a small enough number that my brain is able to remember where they are in the stacking order so Alt+Tab works nicely for switching. Actually Alt+Tab works pretty well with the alternate-tab extension. Except that closing windows, launching of new windows and switching virtual desktops seems to throw everything off.

I use Alt+Tab only for frequent switching between two apps. For everything else, especially for such a small number of windows as 3 I use "Super" key to get to the Activities overview and then I use arrow keys to navigate between windows. No mouse needed and it's pretty quick. Closing window: Ctrl+Q Launching a new window of e.g. Nautilus: "Super"-> typing "nau" -> Ctrl +Enter

There are also tons of extensions that provide different ways to launch and switch between apps such as binding shortcuts to certains apps, but IMHO it's for niche use cases, not really something we need to include in default installation.

But no, having to type part of the name of a window takes significantly more time and requires significantly more hand movement (my left hand usually lives right over the Alt and Tab keys, which also happens to be near Alt+F4 and Ctrl+D which takes care of closing most windows). And also requires a lot more brain power, since if I have an editor, a terminal and a web browser open, I already subconsciously know how many times I need to hit the Tab key before I will end up on the window that I want, but thinking of the name of the program is a lot more work.

...

And no you do *not* need a touchscreen for anything. Repeating that does not make it true.

My comments about touchscreens were more about the large amounts of empty space in some places (makes it easier to poke things with a finger) and the drag-to-remove-screen-blanker (which is ONLY helpful if you have a touchscreen).

...

...

But there are a bunch of other "polish" things that I think also need to be fixed. For example:

• There are at least half a dozen Gnome Bugzillas that I've come across

around focus and window stacking issues - and I think I've encountered every one of them. This is maddening. For someone who uses the keyboard as much as possible, this makes the system almost unusable

What are those focus and stacking issues? You have to name them otherwise we can't discuss / solve them.

I'll start a new thread.

...

...

• The default theme makes it difficult to easily distinguish the Focused

window from all other windows - again, maddening when you have lots of windows spread across multiple very high resolution monitors. Especially with all the window focus/stacking issues

Really? At least for gtk3 apps the whole unfocsued window dims ("backdrop") which makes it very cleary shown as unfocsued. Again no idea what your "focus / stacking issues are"

I'm not saying there is no distinction, I'm saying that it's not a very big distinction (i.e. there needs to be more contrast) so it can be hard to pick out the difference when you have a lot of screen real estate in front of you (i.e. multiple large high resolution monitors) and many windows spread around. If I remember correctly, both KDE's default theme and Windows Aero suffer from the same problems. But I'm trying to use the out-of-the-box defaults since I know they will be better supported (i.e. cross-toolkit theming, support for new widgets, less buggy, etc...) whereas in KDE (which is already non-default) I would have changed the theme and in Windows I would have disabled Aero (active windows have a blue titlebar, inactive get gray, which is fantastic).

-Adam Batkin

...

I think thinking bigger than the above polish items would serve us greatly. I have tons of things that annoy me as well [as a web developer], but there must be bigger fish to fry out there. What would it take for your friends who are developers to move over from a non-free Operating System to Fedora, apart from window management behaviour?

• Andreas

Thanks Andreas - well said, and the right question to ask.

There was a section of the Workstation PRD that mentioned research and user-surveys. I understand that everyone here, especially developers and maintainers, are extremely busy people. I would then suggest the creation of a small research team dedicated to exploring these questions and collecting data for analysis.

Such data would be extremely valuable to the Workstation WG and this type of research would help to begin answering these questions in a meaningful and objective way.

If anyone is interested in forming a Workstation Research group please let me know and I would like to discuss this further...

On Fri, Aug 15, 2014 at 12:22 PM, Chris Murphy lists@colorremedies.com wrote:

On Aug 15, 2014, at 6:19 AM, Adam Batkin adam@batkin.net wrote:

...

I'm not saying there is no distinction, I'm saying that it's not a very

big distinction (i.e. there needs to be more contrast) so it can be hard to pick out the difference when you have a lot of screen real estate in front of you (i.e. multiple large high resolution monitors) and many windows spread around.

I agree. If there's an upstream thread or bug, I'll add a mix of color geek opinion and observed facts about this. I'd slot the problem as "non-death by 1000 cuts" in that the problem seems really minor like it's just a polish concern, and not a reason to not use GNOME, but it probably does adversely impact workflow efficiency.

Chris Murphy

-- desktop mailing list desktop@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/desktop

Matthew Miller píše v Pá 15. 08. 2014 v 10:00 -0400:

On Fri, Aug 15, 2014 at 11:01:26AM +0200, drago01 wrote:

...

You do *not* need to drag it. You can just type you password. Or hit ESC or press Enter. Maybe it is not clear enough that you can do that though.

This is something I've repeatedly seen users struggle with. One particular pain point is that shift/alt/ctrl do not clear the shield -- this is contrary to the training/habits of people who have been using screensavers for a long time, since it's "best practice" to use these keys to reduce the risk of accidentally sending a meaningful key to an app below when the screensaver clears.

I can confirm this. My mother once called me that her system was completely ruined because all she had was a big clock on the screen. I wondered for a while what she did with it again and then I realized that it was a screensaver/lockscreen. She was completely stuck. I have disabled (by an extension) the curtain on my home computer where I don't lock the screen because in this use case I find the curtain useless and it's an additional and unnecessary step to go back to the session.

I also think it would significantly simply things if the unlock-screen password field were just _on_ that screen instead of conceptually hidden by it.

I'm quite opposed to it. The curtain may contain a lot of info (music player controller, all kinds of notifications,...), so it may get cluttered with the password field. But I think the curtain should be lifted by other keys such as Ctrl or Alt and should not be used if the screen is not locked.

Jiri

On Thu, 2014-08-14 at 23:37 -0400, Adam Batkin wrote:

• The default theme uses a lot of extra empty space around UI

controls and also as part of window titlebars. Sure, it's pretty, but you lose a TON of useful space. Open something like Eclipse on a Mac and compare with Gnome and see how much extra "stuff" you can see on the Mac (hint: Apple has found a way to make the UI both pretty AND functional even on screens with limited vertical pixels).

This is a valid point. Adwaita only works well on high-resolution monitors. KDE is able to fit way more functionality into the same amount of space, while still looking good.

Oh, and when I tried another theme that had less blank space around Window titlebars, it turned

out

that some windows use some new Gnome API which draws the titlebars differently, so those windows still look like they are using Adwaita

Sounds like a theme that was never updated for GTK+ 3.10, or not updated properly. Header bars can be themed. The only challenge is that your WM theme must match your GTK+ theme, otherwise they will look out of place next to the WM's decorations used by other apps.

On Tue, Aug 19, 2014 at 04:37:23AM -0400, Christian Schaller wrote:

I don't think this is true. My laptop doesn't have a high-resolution monitor and Adwaita works well. That said with the move to Headerbars there amount of wasted space should become tiny: http://blogs.gnome.org/mclasen/2014/01/13/client-side-decorations-continued/

Some of this is a matter of preference. On my 1366x768 travel laptop, the vertical whitespace used on those headerbars is a lot more than I'd put up with if I had the choice: I'd squeeze the bar down to the height of the buttons, and the buttons themselves to much closer to the height of the text and icons they contain. Those icons/text look to be about 12 pixels high, and the entire bar 48 pixels high — 4× what is necessary.

On a monitor with lots of space, this looks nice. On a constrained display, it feels wasteful — even if the minimal version would be uglier and harder to get used to, it would be nice to have as an option.

On 19/08/14 11:37, Christian Schaller wrote:

I don't think this is true. My laptop doesn't have a high-resolution monitor and Adwaita works well. That said with the move to Headerbars there amount of wasted space should become tiny: http://blogs.gnome.org/mclasen/2014/01/13/client-side-decorations-continued/

Most likely this will work with GTK3 apps, but not with GTK2 or Qt* apps....

On Fri, 2014-08-15 at 16:18 +0200, Andreas Nilsson wrote:

I think thinking bigger than the above polish items would serve us greatly. I have tons of things that annoy me as well [as a web developer], but there must be bigger fish to fry out there. What would it take for your friends who are developers to move over from a non-free Operating System to Fedora, apart from window management behaviour?

• Andreas

Thanks Andreas - well said, and the right question to ask.

...

...

What would it take for your friends who are developers to move over from a non-free Operating System to Fedora, apart from window management behaviour?

• Andreas

Thanks Andreas - well said, and the right question to ask.

On the other hand, it often is "death by thousand cuts" situation. A lot of Mac users would tell you that the reason why they use OS X are all the tiny details that make it more friendly to them. Many of them don't even have a significant concrete reason why they use it, except it's easier and more convenient to use. So little details in UI do matter. But I agree if there are big gaps to fill we should focus on them.

I personally think this is the right line of questioning. But people have been asking this for years. The thing is, now that there is a Fedora "Workstation" focus (i.e. devs, people who are theoretically very comfortable with Linux since everyone deploys to Linux) the question becomes a lot more interesting.

I have a Mac laptop and a Fedora desktop, so I have spent a fair amount of time on both. I find developing on Linux to be much more enjoyable. I would always chose Linux over Mac, except that the Linux laptop experience isn't quite there (I'm being polite). For example, as I mentioned, I can see a lot more on my Mac screen running OS X than Linux, because there is less use of gratuitous white space (sure, some apps have white space, but the OS/theme as a whole is pretty tight - look at the menubar, titlebars, window borders for example). Although I disagree with some Mac design and User Interaction choices, at least there the interface is consistent and follows its own rules. And I won't get started with wireless (which definitely gets better every day), power and thermal management and connecting/disconnecting external displays/projectors.

So why did all of my friends actively switch from Linux to Mac (admittedly, about 10 years ago)? As Alex suggested, I think that's something that needs some research. I would be up for working on that.

But here's the thing: All of the developer tool innovation going on is great, but that's not why people pick Mac over Linux. Mac users have hacked clones of the Linux package managers (MacPorts, Fink, Homebrew), they use Docker via VirtualBox VMs of actual Linux systems, and pretty much everyone ends up deploying to Linux anyway. There may be cool whiz bang developer oriented tools for OS X, but that's because that's where the developers are (i.e. reality/market forces) not because they can't easily build those same tools for Linux. Heck, look at the Atom text editor that GitHub is building - pretty much all of its component parts were originally built *FOR* Linux, and it can *theoretically* be built for Linux, but GitHub doesn't even bother linking to Linux binaries because...well whatever (but of course you can get both Mac and Windows binaries). So there's a case where a developer tool should be available on Linux with a trivial amount of work, and yet its developer still doesn't care.

My theory is that people went Mac for two reasons: 1) Laptop hardware/software Just Works. Maybe Linux on the desktop was difficult back then in the stone age, but everything is fine now. But really, much of the world (even devs) went to laptops. 2) Mac UI/UX Just Works. Even though plenty of people don't agree with Apple's choices, it still works and you know/learn the rules, so it's consistent.

Maybe things like the iTunes Music Store and iPod sync were important, but less so these days (except iOS devs which will always be a problem with any alternate OS). It also helps that if your Mac breaks, you bring it to an Apple store, plop it on the table, wander around the mall for an hour, and come back to a working computer (I'm simplifying, but you get the idea) - and there's no arguing with an incompetent OEM about why some leenuxy thing shouldn't void your warranty and isn't the cause of the scratching sound coming from the hard drive.

So if I had infinite resources, here's probably would I would want to do: * Try to figure out why people went Mac, and what would bring them back (easy peasy, right?) * Make Fedora Workstation work great on laptops. Maybe pick a few specific models and work with an OEM (even if the OEM won't support or sell it, maybe Red Hat has some contacts that can just ensure that we know what the BIOS is doing and what hardware can be expected in a certain model) * Make the UI/UX consistent, clear, intuitive (this is where I think we lose from "death by 1000 cuts" - it's not bad, it's just not...quite there)

-Adam Batkin

On 08/17/2014 03:25 AM, drago01 wrote:

On Sun, Aug 17, 2014 at 5:27 AM, Adam Batkin adam@batkin.net wrote:

...

[..] and connecting/disconnecting external displays/projectors.

Huh? This can't be any easier really. You plug it in .. it works. The only thing we can do here is to make it plug in the cable for you but we lack hardware for that ;)

Maybe life is better these days - I'll be honest, I haven't run Fedora (or any Linux) on a laptop in a few years now.

However (and this is basically the same problem): On my dual-head desktop PC, when I launch a full-screen game, the game is fine (one screen shuts off and the other plays full screen, which is what I want/expect) but when I go back to the desktop, any windows that were on the "secondary" screen have been moved to the "primary" screen. I [think I] know why this happens, and I even have a workaround ("Return to monitor" shell extension) but this raises two questions:

1) Why did I have to go out and find, then install this extension? Who *wouldn't* want this behavior as their default?

2) The extension *seems* to work well, but behind the scenes, I know that bad things (i.e. things I don't want/expect) are happening to my session, and who knows what might be adjusted or reset that I'm just not noticing right now

FYI, that behavior is with Urban Terror (ioquake3). Steam is different - it doesn't mess with my resolution/desktops, but it also doesn't blank the secondary screen which is its own problem. Steam also doesn't seem to let me full-screen its menu system on my primary display, it always fullscreens to the secondary, even though games will launch fullscreen on the primary (and not blank the secondary).

And those are two of the more popular game engines. Maybe it's a bug in the games. But if those two engines, which are pretty popular, can't get it right, maybe a little work is needed on the platform side?

Maybe Wayland will fix all of these things, but Wayland isn't really here yet, plus I have an nvidia card with the closed-source driver, so my understanding is that I'm out of luck for a little while anyway.

I know we're talking "Workstation" not "Desktop" but I also know that one of the goals is that users should be able to do their general computing there too. So unlike the unwashed masses, I'm happy to go through a little extra trouble to make it work, but I'd like the end result to be that it works.

-Adam Batkin

On 08/17/2014 11:05 PM, Alex G.S. wrote:

It's clear that there are some different viewpoints about what should happen. But like I've been saying without independent data this discussion is pointless. Until the Workstation WG is willing to engage in exhaustive market research and user-surveys we will talk in circles.

Okay. Like I said, count me in. How do we do that? I certainly don't have any contacts to do any of the things you suggested. I can sample a half dozen people who use Macs as their primary development platform, but that's a different direction.

But for "Workstation", I think of "developer" not what Dell or HP would call a "Workstation". There are millions of dev/ops people out there with boring old hardware, and that's what I think we want to target. They use lots of terminals, text editors, database tools, etc..Then they bring their laptop home and watch Netflix and play Quake (or whatever).

If the Workstation WG group intends on targeting developers using Windows and Mac OS X they have to take into consideration the fact that these developers are used to a very traditional workstation workflow and are unlikely to accept the alien and highly experimental desktop designs in use on Fedora today.

I agree 100%. I've been kinda beating around the bush in order to be polite. To be honest, I'm used to (and pretty happy with) XFCE, but I'm forcing myself to use Gnome since that's the direction we're (Fedora) trying to head in, and I know that everything will be supported better. The less customization I have to do to be happy, the better. And I suspect that I'm not alone. Gnome (with default theme) will ensure that all widget toolkits are themed the same, that HiDPI will work if I ever get fancy new screens, that I have the latest-and-greatest things like NetworkManager and in general it's better tested (under Fedora).

-Adam Batkin

On 08/17/2014 03:25 AM, drago01 wrote:

On Sun, Aug 17, 2014 at 5:27 AM, Adam Batkin adam@batkin.net wrote:

...

[..] and connecting/disconnecting external displays/projectors.

Huh? This can't be any easier really. You plug it in .. it works. The only thing we can do here is to make it plug in the cable for you but we lack hardware for that ;)

That being said, there are a number of known multi-monitor user experience issues that persist in the latest stable Fedora. These definitely affect Adam's projector use case when you're not in mirroring mode.

https://bugzilla.gnome.org/show_bug.cgi?id=676599 https://bugzilla.gnome.org/show_bug.cgi?id=653085 https://bugzilla.gnome.org/show_bug.cgi?id=668876 https://bugzilla.gnome.org/show_bug.cgi?id=702728 (I think this was fixed recently but not recently enough to be in my version of Fedora)

There are some other issues too I can't find the BZs for atm. Mostly around particular configurations of monitors (primary on the right, vertical stacking, multimonitor workspace issues.)

But this is an upside-down way of going about this, isn't it? How high a priority are external display / multi-monitor concerns in the context of a broader set of developer use cases?

Maybe more along the lines of what Andreas was trying to get at: What are the more broad developer use cases that are important to the developers who've gone OS X?

If we start with the most important use cases, then when we drill down into the minutia, we at least know that fixing those minutia will have a bigger impact on the developer experience than any random bugs that might affect developers. We can use those cases to write usability test scripts that we can run against the latest development version of Fedora and see what UX bugs the developers testing it find. The output of this process would be a prioritized list of papercut style usability issues prioritized by developer use case. Some may be outright bugs requiring development work alone, some may be more complex issues in need of UX designer input.

~m

On 08/19/2014 04:47 PM, Bastien Nocera wrote:

... In that particular case, I'm not sure that it's only a matter of designer input, but rather of manpower for developers initiated in the arcane of window management.

I've been looking through a lot of bugs (both on the Gnome BZ and from my own use) involving multiple monitors, workspaces, window stacking and focus. I think I know what the problem is: There is no specification. No one has fully thought out how it *should* work, so there's no way to tell if something is working properly. Is something a bug, or just how it happens to work, and most people are satisfied enough (or know that complaining is more work than dealing with an inconvenient UX)?

If you wander through the Gnome BZ you will see a ton of things like "I think it makes sense that the window focus should do X when you close a window like this..." and someone else saying "okay here's a patch". This stuff is complex enough that it shouldn't just evolve over time through a series of Bugzilla tickets. And you should be able to understand the [intended] behavior without having to read all of the code or look through the history of every BZ.

Proposal: Let's come up with a specification ("list of rules") for how "we" would like these things to work (window stacking and focus, and how that interacts with multiple monitors and multiple workspaces). This includes different configuration parameters - both things exposed through a UI somewhere and things that may be hidden - that can affect the behavior.

Once that's done, it will be possible to see if the implementation matches what was intended, and if there's a bug, it can be fixed more easily than someone posting a random patch to a BZ saying "hey I came up with a better behavior for X" and everyone trying to figure out how that fits into the bigger scheme of things.

(why Fedora Workstation and not Gnome upstream for this conversation? IMHO our goals are quite different than Gnome's, plus I suspect that there are resources lurking around here that may have the bandwidth and/or skills to accomplish some of these things and further Fedora and Red Hat's goals)

-Adam Batkin

On Tue, Aug 19, 2014 at 11:51 PM, Adam Batkin adam@batkin.net wrote:

I agree that debating multi-screen setups sounds like nitpicking, but the larger point I was making was that Linux on a laptop was (is?) not nearly as pleasant as OS X (or even Windows these days). I like FLOSS but I also like my job, so my laptop needs to be able to connect to a projector (and wifi, and wake up from sleep, etc...) 100% of the time so I'm not fumbling during a presentation. So I have a Mac laptop (running OS X - that's the key, I don't care what kind of hardware I have, but I know that Mac + OS X works).

You do care. Because Thinkpad + OsX wouldn't work :) So you picked very specific hardware.

I think it's unfair to compare Fedora's performance on Macbooks against an operating system that it's designed to work exclusively on this hardware. I have a Thinkpad and all the things you mention (projector, wifi, suspend, etc) work out of the box with Fedora.

~nikos

While we can make our Mac hardware story better, we need to keep in mind that Apple is a pretty hostile company here in terms of running alternative operating systems. There is absolutely no information from them on their hardware, so we are often left with having to reverse engineer to fix bugs, which is slow and time consuming.

My goal here instead is for us to work with more friendly hardware makers to ensure that their hardware work well, so that we can provide Fedora users with a recommendation, i.e. if you want things to work perfectly out of the box get a laptop from vendor X of series Y.

This doesn't mean we shouldn't try to work as well as we can with as much hardware as possible, but realistically we are not in a position to guarantee more than a select set of series.

Christian

----- Original Message -----

From: "Chris Murphy" lists@colorremedies.com To: "Discussions about development for the Fedora desktop" desktop@lists.fedoraproject.org Sent: Friday, August 15, 2014 7:26:11 PM Subject: Re: Developer focus for Fedora workstation

On Aug 15, 2014, at 8:18 AM, Andreas Nilsson lists@andreasn.se wrote:

...

On 08/15/2014 05:37 AM, Adam Batkin wrote:

...

What do people think? I'm trying to be constructive here. Do other people agree or disagree with these ideas? I do have some thoughts on how to actually fix some of these things which might be suited to fresh threads, if I'm not barking up the wrong tree...

I think thinking bigger than the above polish items would serve us greatly. I have tons of things that annoy me as well [as a web developer], but there must be bigger fish to fry out there.

It's a good point.

...

What would it take for your friends who are developers to move over from a non-free Operating System to Fedora, apart from window management behaviour?

I wonder if that question is a trap because it really isn't all that much about the desktop behavior. I use Macs, and there are several show stopper bugs that come well before GNOME or KDE or xfce: overheating and MCE errors, possibly one dead Mac as a result of this; wireless doesn't work out of the box, is a PITA to figure out how to get it to work; trackpad behavior is too erratic to tolerate more than a couple hours worth; I've never had success getting bluetooth to work, thus no alternatives to the trackpad.

Any non-free OS user coming over to Fedora would have their own list of show stoppers before the desktop even comes up. Probably top on the list is video drivers.

How do I get dev friends to move to Fedora when I can't even do that? I'm increasingly likely to give up using Macs to run Fedora, and get hardware that will work instead. My dev friends won't do that. OS X doesn't lend itself all that well to virtualizing other OS's: all the VM's are memory/CPU pigs, and only VirtualBox is FOSS.

So I think the question is a trap because it emphasizes conversion, rather than growing the market for FOSS, and Fedora in particular. What happens if any or all of the following were to happen?

• Make VirtualBox suck less on OS X and Windows, because it seems like a lot

more work getting linux, Fedora in particular, to run; at least on Macs.

• Fedora Workstation runs in a container on Fedora Server, and any OS X,

Windows, Chromebook client uses Fedora Workstation via Guacamole.

• Making it possible and easy to install a Fedora development environment on

OS X (via Macports) and/or Windows. Make the Fedora development environment a portable infiltrating product.

Chris Murphy

desktop mailing list desktop@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/desktop

On Tue, Aug 19, 2014 at 12:16 PM, Chris Murphy lists@colorremedies.com wrote:

At least, Virtual Box should be an explicitly targeted and supported platform to run Fedora on (as a guest). Despite its drawbacks, it must be easier to target than hundreds of models, let alone highly closed and quirky hardware like Macs.

If that is ever going to happen, it needs people interested in it to actually just make it happen. So if anyone would like to see it work well in VirtualBox, then by all means go make it work well in VirtualBox.

We need people to show up and do the work, not tell other people what should be important. We have no lack of the latter.

josh

On Tue, Aug 19, 2014 at 2:48 PM, Chris Murphy lists@colorremedies.com wrote:

On Aug 19, 2014, at 10:19 AM, Josh Boyer jwboyer@fedoraproject.org wrote:

...

On Tue, Aug 19, 2014 at 12:16 PM, Chris Murphy lists@colorremedies.com wrote:

...

At least, Virtual Box should be an explicitly targeted and supported platform to run Fedora on (as a guest). Despite its drawbacks, it must be easier to target than hundreds of models, let alone highly closed and quirky hardware like Macs.

If that is ever going to happen, it needs people interested in it to actually just make it happen. So if anyone would like to see it work well in VirtualBox, then by all means go make it work well in VirtualBox.

We need people to show up and do the work, not tell other people what should be important. We have no lack of the latter.

Convincing people who tell others what is or should be important, are effective recruiters and an incentive for other people to show up and do the necessary work. The inverse is not true.

Really? People have been telling Fedora that virtualbox is very important since it first came out. Apparently the "recruiting" of people that actually want to see it work well is failing. I don't see that being effective at all. At the end of the day you need people that actually DO, not people that just talk.

josh

On Tue, Aug 19, 2014 at 3:59 PM, Chris Murphy lists@colorremedies.com wrote:

On Aug 19, 2014, at 12:56 PM, Josh Boyer jwboyer@fedoraproject.org wrote:

...

On Tue, Aug 19, 2014 at 2:48 PM, Chris Murphy lists@colorremedies.com wrote:

...

On Aug 19, 2014, at 10:19 AM, Josh Boyer jwboyer@fedoraproject.org wrote:

...

On Tue, Aug 19, 2014 at 12:16 PM, Chris Murphy lists@colorremedies.com wrote:

...

At least, Virtual Box should be an explicitly targeted and supported platform to run Fedora on (as a guest). Despite its drawbacks, it must be easier to target than hundreds of models, let alone highly closed and quirky hardware like Macs.

If that is ever going to happen, it needs people interested in it to actually just make it happen. So if anyone would like to see it work well in VirtualBox, then by all means go make it work well in VirtualBox.

We need people to show up and do the work, not tell other people what should be important. We have no lack of the latter.

Convincing people who tell others what is or should be important, are effective recruiters and an incentive for other people to show up and do the necessary work. The inverse is not true.

Really? People have been telling Fedora that virtualbox is very important since it first came out. Apparently the "recruiting" of people that actually want to see it work well is failing.

No it's either because it's a bad idea to begin with, or there's no one trusted with the cachet to sell the idea, or that person hasn't himself been convinced.

Convincing people = people who are convincing, could be one person, they're typically in leadership roles. Convincing people ≠ a volume of opinions.

The volume of opinions is noise, the doers aren't listening, or don't have a reason to care. Too many cats, too few cat herders.

How is any of that not considered failing to promote Fedora+VirtualBox? I honestly have no idea what your point is here.

...

I don't see that being effective at all. At the end of the day you need people that actually DO, not people that just talk.

I'd characterize the project's attitude toward Virtual Box as somewhere in the realm of anemic, not quite openly hostile. So it doesn't surprise me it's not attracting Fedora+VirtualBox specific developers to make the combination a better overall experience.

Great. I'd agree with all of that. So I'm not sure why you think talking about it more is going to make it more important, or why all of a sudden now people would start digging in.

josh

On Tue, Aug 19, 2014 at 7:27 PM, Chris Murphy lists@colorremedies.com wrote:

On Aug 19, 2014, at 2:06 PM, Josh Boyer jwboyer@fedoraproject.org wrote:

...

On Tue, Aug 19, 2014 at 3:59 PM, Chris Murphy lists@colorremedies.com wrote:

...

On Aug 19, 2014, at 12:56 PM, Josh Boyer jwboyer@fedoraproject.org wrote:

...

On Tue, Aug 19, 2014 at 2:48 PM, Chris Murphy lists@colorremedies.com wrote:

...

On Aug 19, 2014, at 10:19 AM, Josh Boyer jwboyer@fedoraproject.org wrote:

...

On Tue, Aug 19, 2014 at 12:16 PM, Chris Murphy lists@colorremedies.com wrote: > > At least, Virtual Box should be an explicitly targeted and supported platform to run Fedora on (as a guest). Despite its drawbacks, it must be easier to target than hundreds of models, let alone highly closed and quirky hardware like Macs.

If that is ever going to happen, it needs people interested in it to actually just make it happen. So if anyone would like to see it work well in VirtualBox, then by all means go make it work well in VirtualBox.

We need people to show up and do the work, not tell other people what should be important. We have no lack of the latter.

Convincing people who tell others what is or should be important, are effective recruiters and an incentive for other people to show up and do the necessary work. The inverse is not true.

Really? People have been telling Fedora that virtualbox is very important since it first came out. Apparently the "recruiting" of people that actually want to see it work well is failing.

No it's either because it's a bad idea to begin with, or there's no one trusted with the cachet to sell the idea, or that person hasn't himself been convinced.

Convincing people = people who are convincing, could be one person, they're typically in leadership roles. Convincing people ≠ a volume of opinions.

The volume of opinions is noise, the doers aren't listening, or don't have a reason to care. Too many cats, too few cat herders.

How is any of that not considered failing to promote Fedora+VirtualBox? I honestly have no idea what your point is here.

I am refusing the premise that some vocal users asking for a better experience constitutes promoting this combination.

There has been no Fedora leadership at any level acknowledging this is a good, valid, promotable idea. Nor that it's a an inherently flawed one, fraught with peril, and should be abandoned once and for all. The current status is one of ambivalence. It's oozes "we don't care, if you care then go do that," with maybe a good luck cherry on top.

It oozes that because that's exactly what it is. Let's stop agreeing on this point and get to something relevant. I honestly have no idea if you're trying to advocate it shouldn't be, but if you are you're doing it in a very confusing manner.

In contrast, I see Server and Cloud products having more clearly benefited from their self-imposed constraints. Yet they still have a "build it and they will come" attitude.

More and more often I'm seeing emails pointing fingers at OS X, making comparisons. I'm not the only one observing [1] the prevalence of Macs at FOSS conferences that are not running a FOSS OS but rather OS X. And I'm wondering what Workstation wants to be when it grows up and if it needs additional self-imposed constraints to help it along.

Anything built in the past 8 years is recommended hardware, but none of it's explicitly supported as far as I can tell.

Welcome to community driven, volunteer supported Linux distributions... We can't tell people with new hardware "sorry, we don't support you because you didn't buy a thinkpad or the $vendor of the week." We're losing enough users already. We also can't say "this is explicitly supported" because the distro and the hardware is constantly changing, and they aren't changing in lock-step or with any coordination between the distro and the HW vendors. So we do as best we can.

Can Fedora Workstation be viable as predominately a VM guest on some other OS? Can this be a good experience? What resources are required to make that happen?

If someone works on it. If they do, they can figure out what they need.

What's the alternative?

Uh, I kind of view Fedora working on some _other_ OS as the alternative. Which is probably the case for a lot of people, which is probably why nobody is really driving it.

What's the autopsy report on the various Mac hardware problems?

You have a Mac with problems. I know at least 7 other people that have Macs that work fine, with the exception of the 802.11n issue which we can't really fix.

I can only describe symptoms, the bugs are all essentially unanswered as if no one has any concrete idea what the underlying problem actually is, therefore I don't know if it's worth throwing resources at it?

Yep. That's also par for the course. People basically have to guess at a lot of the problems you've described if they can't recreate on their own hardware. I'm beginning to wonder if you're confusing Fedora with some kind of paid-support model distribution.

What's the alternative to that? Should we put together a list of actually recommended hardware, specific makes and models? And if so what are the (largely) objective criteria by which to figure out what is recommended?

We can't do that without some kind of relationship with vendors for that hardware. Otherwise we're playing catch up at best trying to make things work _after_ the hardware has shipped and it really doesn't change the user experience or support story from how things are today. Christian has mentioned trying to build such relationships, which would be good. It won't be with Apple.

...

...

...

I don't see that being effective at all. At the end of the day you need people that actually DO, not people that just talk.

I'd characterize the project's attitude toward Virtual Box as somewhere in the realm of anemic, not quite openly hostile. So it doesn't surprise me it's not attracting Fedora+VirtualBox specific developers to make the combination a better overall experience.

Great. I'd agree with all of that. So I'm not sure why you think talking about it more is going to make it more important, or why all of a sudden now people would start digging in.

New project focus, new product name, seems like a good time for a change in tone. The project's attitude toward Virtual Box would have to change before anything else possibly could.

Again, not sure if you're advocating for it to change. If you are, you're doing it in a confusing manner.

josh

I don't want to sound selfish, but if I have these issues, there are probably others who either don't want to speak up or haven't fully articulated what it is that isn't right - or have moved to KDE or XFCE and never looked back, and we should try to get them back too...if your core dedicated userbase is scattered among DEs, how do you expect to bring people in who aren't so committed?

Fedora is a distribution and it doesn't compete directly against GNOME, KDE or LXDE or any desktop environment. Fedora Workstation will compete against Ubuntu, Arch and OpenSUSE as well as Apple's OS X for developer mindshare and contributors. In order to succeed it has to appeal to a wide general audience of diverse users/developers with very diverse tastes and workflows.

GNOME is a desktop environment that competes against KDE, LXDE and any other desktop environment out there. It is available on Fedora's competitors Ubuntu, Arch and OpenSUSE as well as many other distributions. GNOME, the desktop environment, has as it's goal, like any desktop enviornment, to appeal to a certain subset of users who happen to like it's workflow, it is not a general purpose project but specific.

The interests of GNOME and Fedora Workstation are totally different and because GNOME can be run on Fedora's competitors these interests aren't always in alignment.

Let's say Fedora puts all this investment in time and money into improving GNOME with the Fedora Workstation project. What's stopping these changes from propagating to Fedora's competitors? Nothing. GNOME users on Ubuntu, Arch and OpenSUSE will also have the full "Fedora Workstation" experience. There's no need to actually switch to Fedora and thus Fedora will see no net-growth in it's user-share or new developers. Instead the effort of Fedora's developers will go to help grow Ubuntu with it's GNOME spin.

If the Fedora distribution was to shutdown and disappear tomorrow GNOME would still be available on Ubuntu, Arch and OpenSUSE and going strong.

What should Fedora do to ensure that Fedora Workstation is successful?

Fedora Workstation's real value will be how it integrates developer tools with multiple desktop environments!

Focus on Fedora Workstation as a platform and integrate it with a series of mainstream desktop environments such as GNOME, KDE and LXQt. Make sure these environments are deeply integrated with the whole Workstation experience so that users/developers get a unique enterprise-level experience they can't get on Ubuntu, Arch or OpenSUSE. When the user installs Fedora Workstation present them with a series of desktop choices to install.

In reality you shouldn't care what desktop the user/developer is using, you should care what distribution they're using and that's the whole point...

On Tue, Aug 19, 2014 at 10:08 PM, Adam Batkin adam@batkin.net wrote:

On 08/19/2014 09:10 PM, Josh Boyer wrote:

...

In contrast, I see Server and Cloud products having more clearly

...

benefited from their self-imposed constraints. Yet they still have a "build it and they will come" attitude.

More and more often I'm seeing emails pointing fingers at OS X, making comparisons. I'm not the only one observing [1] the prevalence of Macs at FOSS conferences that are not running a FOSS OS but rather OS X. And I'm wondering what Workstation wants to be when it grows up and if it needs additional self-imposed constraints to help it along.

Anything built in the past 8 years is recommended hardware, but none of it's explicitly supported as far as I can tell.

Welcome to community driven, volunteer supported Linux distributions... We can't tell people with new hardware "sorry, we don't support you because you didn't buy a thinkpad or the $vendor of the week." We're losing enough users already. We also can't say "this is explicitly supported" because the distro and the hardware is constantly changing, and they aren't changing in lock-step or with any coordination between the distro and the HW vendors. So we do as best we can.

...

I can only describe symptoms, the bugs are all essentially unanswered as

...

...

if no one has any concrete idea what the underlying problem actually is, therefore I don't know if it's worth throwing resources at it?

Yep. That's also par for the course. People basically have to guess at a lot of the problems you've described if they can't recreate on their own hardware. I'm beginning to wonder if you're confusing Fedora with some kind of paid-support model distribution.

What's the alternative to that? Should we put together a list of

...

actually recommended hardware, specific makes and models? And if so what are the (largely) objective criteria by which to figure out what is recommended?

We can't do that without some kind of relationship with vendors for that hardware. Otherwise we're playing catch up at best trying to make things work _after_ the hardware has shipped and it really doesn't change the user experience or support story from how things are today. Christian has mentioned trying to build such relationships, which would be good. It won't be with Apple.

Personally, my hope is that some issues will be taken up by Red Hat employees, either because they happen to have an interest in the area, or because Red Hat will make it their responsibility because it's in RH's interest. I believe that it's in Red Hat's best business interest to work on certain things that will enhance Fedora, which will in turn enhance RHEL (even if it's just to stem the tide of people moving to Ubuntu, or bring people who would have otherwise chosen Ubuntu over).

For developer workstations, the competition is mostly Mac. For server deployments, it's Ubuntu. I don't think Red Hat "loses" anything when people use Macs for development (though they might gain if people used RHEL instead) but they *do* lose if people choose Ubuntu for deployment. And if someone chooses Ubuntu for a workstation (say, because they have a better relationship with a hardware vendor) instead of Fedora, that *does* hit Red Hat because that developer may have a greater likelihood of deploying to Ubuntu instead of RHEL.

And for everyone else who does this of their own volition (maybe because Fedora is more ideologically FLOSS): It's in your best interest to make Fedora better for everyone too. If the rest of the world thinks "Linux=Ubuntu" then it will become harder and harder to get things to work nicely under Fedora (and RHEL/Centos) as everyone builds solely for Ubuntu.

New project focus, new product name, seems like a good time for a change

...

...

in tone. The project's attitude toward Virtual Box would have to change before anything else possibly could.

Again, not sure if you're advocating for it to change. If you are, you're doing it in a confusing manner.

Can someone explain what all the fuss is around running Fedora under a Virtual Machine and what's wrong with VirtualBox support?

I run a bunch of Fedora VMs and they all run fine. I even run Fedora under VB on my Mac and it's as good as can be expected (given the Mac keyboard layout and how Mac steals certain key combos). It's slow, but that's probably my 6 year old laptop.

I'd vote that Workstation focus should be on something other concentrating on being a VM guest, but only because my experience there has been pretty good, and most of my complaints affect Fedora whether it's run as a VM or bare metal.

I really think that we need a better consensus on the long term goals of Workstation, since there have been lots of people (myself included) jumping up and saying "let's do this and that". The official mission statement(s) are pretty vague, and while the "Overall plans and policies for the product" is good, I personally think that it's lacking any discussion of UI/UX, and as a long time committed Fedora/Red Hat user, that's the area where I have the most pain right now (I don't want to sound selfish, but if I have these issues, there are probably others who either don't want to speak up or haven't fully articulated what it is that isn't right - or have moved to KDE or XFCE and never looked back, and we should try to get them back too...if your core dedicated userbase is scattered among DEs, how do you expect to bring people in who aren't so committed?).

-Adam Batkin

-- desktop mailing list desktop@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/desktop

On 08/20/2014 06:49 PM, drago01 wrote:

No. The biggest competitor is not Mac but Windows it dies not only have a way bigger marketshare but also trys to support a wide range of hardware (like us) the only difference is that is has better support form hardware vendors. This becomes even more true if you include the non US market. So stop focusing to much on Apple there are other (bigger) competitors out there.

Is Windows really our competition?

I see Windows in "Microsoft shops" (i.e. Windows servers) and in "corporate" environments where I don't think Linux (or Mac) has a chance.

The Fedora Workstation PRD defines 4 specific cases (Student, Independent Developer, Small Company Developer and Developer in a Large Organization) and I would wager that for cases 1, 2 and 3 the competition is largely Mac and to a smaller extent, Ubuntu.

Still, I think most of what I've been throwing out there is applicable whether the "competition" is Mac, Windows or Raspberry Pi.

People (rightfully) spend a lot of time comparing with Mac because Mac has risen very quickly from almost non-existence to a disproportionate percentage of the developer community. So what did Apple do correctly to attract so many developers (most of whom are deploying to Linux) that perhaps we can improve upon, to bring them back, or bring new people in?

-Adam Batkin

Adam Batkin wrote:

Personally I think it's useful to consider what OS X and Windows do, and whether Fedora can do better, but if there's any competition it's between Linux distros. If we want more users we should go after Ubuntu and Linux Mint users first.

The problem isn't as simplistic as "let's build a vertically integrated desktop solution" and then hope that there's some magical migration from Mac and Ubuntu Unity or Linux Mint users over to Fedora, it's a more complex problem than it might appear.

Fedora Workstation is about becoming the de-facto reference platform for developers doing commercial and/or important projects, a standard.

But the problem is that currently Fedora isn't that go-to platform, it's Ubuntu:

Example #1:

Let's take the Nvidia Jetson TK1 ARM developer board, this by default runs Ubuntu pre-loaded on the board.

https://developer.nvidia.com/jetson-tk1

Example #2:

Steam for Linux, Valve used Debian for SteamOS and recommends Ubuntu for the desktop.

"Ubuntu is our favorite version of Linux. Interested in giving it a whirl? " --- Valve

http://store.steampowered.com/about/

That's the real problem at hand. Fedora is losing out to Ubuntu as the standard reference platform and not engaging with developers from outside it's own community. This then translates into things like server deployments and lost opportunities on the OEM side. But oddly enough Fedora and it's community are major contributors to the most important and fundamental Linux projects like the kernel, systemd, graphics, OpenStack and more...

That's why I think bothering with the desktop environment layer and making a default experience isn't feasible and doesn't make any sense given the "upstream" nature of Fedora.

Desktop environments make the distribution a replaceable commodity. If you're a GNOME user then you can move from Fedora to Ubuntu to Arch or anything else and still keep using GNOME with minimal disruption. The same applies to any desktop environment including KDE. People moving away from Fedora can do so easily because of this. To get the latest upstream development versions they can just add a PPA in Ubuntu or just install Arch or a derivative like Antergos, they don't need Fedora for that.

That's why I suggest that Fedora developers and contributors interested in the desktop environment layers should form their own independent SIG's (special interest groups) and have that be a layer above what is defined as "Fedora Workstation". That is "Fedora Workstation" should not be a mere "showcase" for a particular desktop project but an expansive platform. There would be no "default" desktop environment at all or even the concept of a default.

If these SIG's want to go ahead and create their own hand-crafted experiences let them do it. Containers could really be an innovative way to allow many different and potentially incompatible desktop experiences to exist on the same system. These things are only possible if you further separate the layers here. It's also wise from a political perspective because you could engage many diverse communities around Fedora Workstation this way. They would feel like equal stakeholders and not feel like second-class citizens.

Embracing the diversity with SIG's is the way to rally the wider developer community around Fedora Workstation.

On Wed, Aug 20, 2014 at 8:13 PM, Adam Batkin adam@batkin.net wrote:

On 08/20/2014 06:49 PM, drago01 wrote:

...

No. The biggest competitor is not Mac but Windows it dies not only have a way bigger marketshare but also trys to support a wide range of hardware (like us) the only difference is that is has better support form hardware vendors. This becomes even more true if you include the non US market. So stop focusing to much on Apple there are other (bigger) competitors out there.

Is Windows really our competition?

I see Windows in "Microsoft shops" (i.e. Windows servers) and in "corporate" environments where I don't think Linux (or Mac) has a chance.

The Fedora Workstation PRD defines 4 specific cases (Student, Independent Developer, Small Company Developer and Developer in a Large Organization) and I would wager that for cases 1, 2 and 3 the competition is largely Mac and to a smaller extent, Ubuntu.

Still, I think most of what I've been throwing out there is applicable whether the "competition" is Mac, Windows or Raspberry Pi.

People (rightfully) spend a lot of time comparing with Mac because Mac has risen very quickly from almost non-existence to a disproportionate percentage of the developer community. So what did Apple do correctly to attract so many developers (most of whom are deploying to Linux) that perhaps we can improve upon, to bring them back, or bring new people in?

-Adam Batkin

-- desktop mailing list desktop@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/desktop

On Aug 21, 2014, at 12:34 PM, Chris Murphy lists@colorremedies.com wrote:

On Aug 19, 2014, at 8:08 PM, Adam Batkin adam@batkin.net wrote:

...

Can someone explain what all the fuss is around running Fedora under a Virtual Machine and what's wrong with VirtualBox support?

Carving out a test criteria that permits certain bugs to be blocking, hence some level of VirtualBox support.

The whole thread is much longer but this is a decent one-shot is this: https://lists.fedoraproject.org/pipermail/test/2012-November/111614.html

Nothing has really changed in 21 months policy wise.

For example if it turns out this bug is in fact VirtualBox specific, what happens? Historically, it's not a blocker expressly because it's happening in vbox. And that would mean F21 would be allowed to ship, despite failing to install as a virtualbox guest. If it's only fixable by virtualbox upstream, it's reasonable to not block because we can't control any of that. If it can be fixed in Fedora, but we need more time or whatever to get it fixed, then I think we should block. But we're not allowed to block on anything vbox specific no matter what the problem or solution is.

https://bugzilla.redhat.com/show_bug.cgi?id=1103496

Chris Murphy

On Aug 21, 2014, at 2:20 PM, Lars Seipel lars.seipel@gmail.com wrote:

On Thu, Aug 21, 2014 at 01:13:07PM -0600, Chris Murphy wrote:

...

If it can be fixed in Fedora, but we need more time or whatever to get it fixed, then I think we should block. But we're not allowed to block on anything vbox specific no matter what the problem or solution is.

https://bugzilla.redhat.com/show_bug.cgi?id=1103496

The problem with that bug doesn't seem to be a lack of time but more a lack of interest. If no one cares enough about the bug to go after it, it won't get magically fixed by just waiting for it, except maybe by coincidence. Blocking the release on it might coerce some contributor into fixing it but that's not a viable long-term strategy for a project.

And never blocking is also not a viable long-terms strategy. The blocking dance is a skill to equalize the fact some problems are important enough even if it's not affecting so many people that it immediately gets fixed. This happens all the time and if it didn't we'd have a much lower quality release.

To support something you really need people interested in working on it. In all the time that bug has been open apparently no one thought about talking to Virtualbox upsream and going through the instructions for debugging such issues. At least the bug shows no evidence of that.

That's misleading, it's only today that someone posted a supposition it *might* be VirtualBox specific. But comment 2 is from one of the anaconda developers who has seen this also, and AFAIK he doesn't use VirtualBox, nor do any anaconda developers that I'm aware of (meanwhile I've also not seen them suggest the slightest hint of disapproval or skepticism of vbox based bugs or testing). So it's uncertain that this is only a VirtualBox bug.

Another way to narrowly carve this would be if "nomodeset" in either BIOS or UEFI mode boot allows it to work, then we shouldn't block even if the problem were to occur with guest additions built (which of course taints the kernel and leads to a more complicated troubleshooting process).

This action has to come from Virtualbox users. Anyone else is unlikely to set up a testing environment (which is a pretty invasive thing to do with vbox and unlikely to work with recent kernels) to work on a bug he or she personally does not care for.

There are two dangerous flaws with this line of logic:

1.

There are a number of QA testers who predominately use VirtualBox for testing Fedora; and they (including me) tick a box in a test matrix to indicate whether a particular test has passed or failed. The suggestion that VirtualBox bugs shouldn't block, but tests that base in VirtualBox should be accepted is a contradiction. Yes thank you I'll take the good result, but the bad result doesn't matter.

2.

We have an ecosystem where many people of different interests and skill sets do different things, and this enriches the ecosystem and makes it more valuable than the individual skill sets alone. People do things that aren't even directly valuable to themselves, but they do it anyway because they're good at it, and it's for the common good. What goes around comes around.

The attitude you are describing is "every man for himself, if you want something done, do it yourself" and distinctly so by singling out the VirtualBox users. This is not common good based. The equivalent would be me saying, OK, screw it, I'm not going to test anything anymore. If it's broken, I'm not filing bugs. I'm not going to try and trouble shoot it. I'll just go do something else, or find an alternative that works right now. I'll let other people do that leg work, because I don't want to do it, and I don't see myself as part of something bigger. It's only about me.

So be careful what you ask for. I really don't think this attitude is going to help with the attrition problem.

Chris Murphy

On Thu, Aug 21, 2014 at 04:17:07PM -0600, Chris Murphy wrote:

The attitude you are describing is "every man for himself, if you want something done, do it yourself" and distinctly so by singling out the VirtualBox users

That's not what I'm saying (and definitely not what I mean). The issue is that the release criteria should correspond to the resources realistically available or otherwise it's just going to lead to frustration. A way to ensure that would be consistently meeting the proposed criteria prior to them being enforced (like it is done with secondary arch promotions).

If something doesn't have enough support in the wider community to demonstrate its feasibility, prescribing its implementation by means of policy isn't doing much good.

Lars

----- Original Message -----

While we can make our Mac hardware story better, we need to keep in mind that Apple is a pretty hostile company here in terms of running alternative operating systems. There is absolutely no information from them on their hardware, so we are often left with having to reverse engineer to fix bugs, which is slow and time consuming.

That's mainly wrong. The hardware from this year's MacBook Pros is 99% the same as last year's MacBook Pros. Apart from the usual rigmarole of keyboard and touchpad USB IDs updates, it works as well as last year's. The only thing that doesn't work out of the box on those machines is the wireless card.

The dual boot experience is sub-par, but that's mainly because we regressed the installer compared to when Matthew fixed all that a couple of years ago.

My goal here instead is for us to work with more friendly hardware makers to ensure that their hardware work well, so that we can provide Fedora users with a recommendation, i.e. if you want things to work perfectly out of the box get a laptop from vendor X of series Y.

While that's a great goal, we're not going get Lenovo (for example) releasing *less* SKUs. With Apple, you get 2 new SKUs per year for the laptops (one Pro, one Air). That's much easier to target.

That's not to say that a brand new funky model from Apple wouldn't be a royal pain to get supported, but the amount of work required these days isn't on par with supporting a new laptop from some of the other makers ("hey! here's a completely new touchpad you don't have a driver for, enjoy!").

This doesn't mean we shouldn't try to work as well as we can with as much hardware as possible, but realistically we are not in a position to guarantee more than a select set of series.

Even if we should privilege hardware manufacturers that will support us, we (that's Fedora and Red Hat as well) should still spend time on making sure Apple's hardware works. (Slightly bogus numbers ahead) If they sell half as many laptops as HP, but HP has 10 new SKUs in the year, I would know which one to focus on to reach more potential users.

Cheers

On Thu, Aug 14, 2014 at 11:37:32PM -0400, Adam Batkin wrote:

• The default theme makes it difficult to easily distinguish the

Focused window from all other windows - again, maddening when you have lots of windows spread across multiple very high resolution monitors. Especially with all the window focus/stacking issues

There is a "shade inactive windows" extension you might find helpful here, although I find that its default shading level is more than I like.

What do people think? I'm trying to be constructive here. Do other people agree or disagree with these ideas? I do have some thoughts on how to actually fix some of these things which might be suited to fresh threads, if I'm not barking up the wrong tree...

I think keeping track of "UI papercuts" is a useful thing. Not everone is going to agree with every one, of course, but having a collective sense is valuable, and I think we *can* put more emphasis on some of the developer use.