Christian Schaller cschalle at redhat.com Thu Apr 2 14:19:55 UTC 2015
I think a lot of these items are things we are already aware of and trying to fix, but of course not all of them are easily fixable, like access to proprietary Windows or MacOS X applications or similar hotkeys/behaviour across UI toolkits. I think we made some great strides in the stability department, but reading the reddit thread did reinforce that it is an area we need to keep focus going forward.
Stability and polish seems challenging with two releases per year, each one with only 13 months of security updates. Conversely on OS X it's one year with software updates, 2-3 years of security updates, and yet new application versions work on at least 2 major OS X versions if not 3 or 4.
Fedora churn emphases newness over stability and polish as features.
Another is whether a Linux distro, including Fedora, is an OS or just a collection of packages? I see more emphasis on components, rather than how it fits into a whole. Apple has its walled garden in the form of proprietary software, but FOSS puts upstreams into walled gardens of their own, essentially immunized from feature requests let alone anything approaching insistence.
In a recent installer password quality thread on the security@ list, mitr writes "All of this just takes a willingness to look at a dozen components at a time instead of at a single one, and a willingness to write patches that sum up to thousand lines instead of a single five-line patch." https://lists.fedoraproject.org/pipermail/security/2015-February/002079.html
And I'd extrapolate that to the entire OS. Are there generalists with the distro's vision, and political capital to inform and convince upstreams in a cooperative manner? The component specialists really don't seem to have this interest, outside of the desktop environments.
I think Fedora should not model itself after, or chase the IDE mentality of Apple or Microsoft. In my view, it isn't really free anyway, just because of a licensing difference. It's only made really free by bridging the already large and widening gap between the application developer and user. The opposite almost happened [1], but those companies abandoned that path. So quite honestly, I'd say, screw these other models of development, all you can hope for replicating them with a different frosting but the substance will still amount to the same thing and for that, why bother?
[1] http://thetrendythings.com/read/20042
I see three weaknesses for us that we must address:
* Marketing: we're seriously underrated compared to other distros, considering the quality of what we release relative to what other distros are releasing. With very important exceptions: * There's no reasonable way to upgrade from one Fedora release to the next. * Post release updates are crazy.
And one weakness we can't really address:
* Legal issues (multimedia)
(Although we should make an effort to get MP3 support in F23, since the last problematic patent we know about expires in September.)
On Thu, 2015-04-02 at 21:29 -0600, Chris Murphy wrote:
Christian Schaller cschalle at redhat.com Thu Apr 2 14:19:55 UTC 2015
I think a lot of these items are things we are already aware of and trying to fix, but of course not all of them are easily fixable, like access to proprietary Windows or MacOS X applications or similar hotkeys/behaviour across UI toolkits. I think we made some great strides in the stability department, but reading the reddit thread did reinforce that it is an area we need to keep focus going forward.
FWIW I think we've already fixed most of those complaints long ago. Anything related to digging in /etc to muck with settings is only true for power users; you don't have to do that on Workstation, unless you're trying to tweak something that you probably do not need to tweak, and would not be able to tweak at all on a Mac. That you have the *option* to dig into text files and use the terminal does not mean you *must* do so. (We need a slogan for that.)
The multimonitor complaints related to Xfce, I think.
In general, we need to differentiate ourselves better from other Linux distributions, because we're being dragged down by people's impressions of other distros, or their impressions of what Fedora was like 10 years ago. "Fedora is a test bed for RHEL" and "Fedora is too hard for new users" are the particularly damaging thoughts, which we must replace with "Workstation just works" or something like that.
Stability and polish seems challenging with two releases per year, each one with only 13 months of security updates. Conversely on OS X it's one year with software updates, 2-3 years of security updates, and yet new application versions work on at least 2 major OS X versions if not 3 or 4.
I think we're better off targeting the new Windows model, where upgrades get pushed semi-regularly as part of the normal updates process. If the upgrade is graphical and safe, we should be fine. The current issue is that there's no way for a normal human being to know when his release is outdated or unsupported, let alone upgrade from one to the next. We need to show a notification with GNOME Software and handle the upgrade there. We have mockups for this already: https://raw.githubusercontent.com/gnome-design-team/gnome-mockups/master/sof...
I'm satisfied with the quality/stability of Fedora that we release. Our QA for releases is significantly better than for any other distro with a comparable release cycle, including Ubuntu (their "installer wipes my entire disk without asking" bug -- did they ever fix that? -- would have been a blocker for us).
I am *not* satisfied with the quality/stability of Fedora once you press the Install Updates button. That is our biggest problem. I want to adopt something closer to Ubuntu's SRU policy (although not as strict as that) for Workstation, where package maintainers do not get to approve their own updates. Most importantly, package maintainers should be required to *justify* updates, and also prepare the changelogs properly using PackageKit markdown, before they get approved for Workstation.
We would use this process not just to sanity check the updates we release, but more importantly to reduce the frequency of updates. I'm less comfortable saying "no you can't fix bugs in your package" than I am saying "no you can't fix bugs in your package every week, wait a bit longer." It's not really reasonable to update most packages more than once per month, IMO.
I'm also interested in pushing non-security all together once per month.
Fedora churn emphases newness over stability and polish as features.
Another is whether a Linux distro, including Fedora, is an OS or just a collection of packages? I see more emphasis on components, rather than how it fits into a whole. Apple has its walled garden in the form of proprietary software, but FOSS puts upstreams into walled gardens of their own, essentially immunized from feature requests let alone anything approaching insistence.
Well, I think it's both: some packages are part of the core OS, and others aren't. Terminal, Files, and gdm are part of the OS. Eclipse is not.
On Fri, Apr 03, 2015 at 09:33:59AM -0500, Michael Catanzaro wrote:
I see three weaknesses for us that we must address:
- Marketing: we're seriously underrated compared to other distros,
considering the quality of what we release relative to what other distros are releasing.
Reviews in the press for F21 were very good. Early reviews for F22 are also good. I think there are lots of *visible* good things happening in Fedora (the redesigned start.fp.o, gnome-software and gnome in general, wallpapers, design polish, etc). One release is not enough, but I think users will follow if we can keep this up for a few releases.
With very important exceptions:
- There's no reasonable way to upgrade from one Fedora release to the
next.
- Post release updates are crazy.
This is unnecessarily bleak. Offline update a la fedup is a technically sound approach and can be made to work reliably. I know there are problems right now, systemd is one of the components that has issues, but it's nothing that can't or won't be solved.
Post release updates, even of large fraction of the package set, should be OK, as long as they are not disruptive. In my subjective feeling, things are getting better here too. There's a fairly tight update policy and at the level of stable updates things mostly work.
And one weakness we can't really address:
- Legal issues (multimedia)
(Although we should make an effort to get MP3 support in F23, since the last problematic patent we know about expires in September.)
MP3 support out of the box would be nice.
Zbyszek
With very important exceptions:
- There's no reasonable way to upgrade from one Fedora release to the
next.
- Post release updates are crazy.
This is unnecessarily bleak. Offline update a la fedup is a technically sound approach and can be made to work reliably.
No update process that lacks a GUI can be considered reasonable. Remember, developers don't want to use terminal (although familiar with it) and commands found on a wiki to update their system.
Post release updates, even of large fraction of the package set, should be OK, as long as they are not disruptive.
If I get prompted any other day to update and reboot it is disruptive.
On Fri, 2015-04-03 at 22:10 +0000, Nikos Roussos wrote:
No update process that lacks a GUI can be considered reasonable. Remember, developers don't want to use terminal (although familiar with it) and commands found on a wiki to update their system.
Yeah, I think we're all on the same page regarding that. It's something we want, but nobody has had time to work on it.
If I get prompted any other day to update and reboot it is disruptive.
I've actually noticed that I'm only getting the update notification about once per week. That's been working well, although I wish the size of the weekly updates were smaller.
But I kind of suspect that PackageKit is not detecting anything as a security update. :)
On Fri, Apr 03, 2015 at 07:11:36PM -0500, Michael Catanzaro wrote:
I've actually noticed that I'm only getting the update notification about once per week. That's been working well, although I wish the size of the weekly updates were smaller.
Something I keep meaning to do is go back through the history of the last few releases and actually graph the number of security updates that affect the default Workstation/Cloud/Server package sets every day. We definitely push security updates very often, but I don't have a great sense for their impact.
desktop@lists.fedoraproject.org
-
Chris Murphy -
Matthew Miller -
Michael Catanzaro -
Nikos Roussos -
Zbigniew Jędrzejewski-Szmek