FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Redhat > Fedora Development

 
 
LinkBack Thread Tools
 
Old 02-26-2010, 12:49 PM
Till Maas
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

On Fri, Feb 26, 2010 at 08:36:41AM -0500, Josh Boyer wrote:
> On Fri, Feb 26, 2010 at 02:23:33PM +0100, Till Maas wrote:
> >On Fri, Feb 26, 2010 at 01:16:43PM +0100, Kevin Kofler wrote:
> >
> >> I would like to collect feedback on this issue. If you want to disable
> >> direct stable pushes, why? Could there be a less radical solution to that
> >> problem (e.g. a policy discouraging direct stable pushes for some specific
> >> types of changes rather than a blanket ban)? On the other hand, if (like me)
> >> you DON'T want that feature to go away, please provide valid use cases.
> >
> >Imho it takes too long to get packages into updates-testing, if people
> >are really interested in testing packages, they often seem to get
> >packages directly from Koji, e.g. on this update I got 3 positive Karma
> >points (one of them was anonymous) within 76 minutes after submitting
> >the update:
> >https://admin.fedoraproject.org/updates/F12/FEDORA-2010-0604
> >
> >It did not seem very useful to delay this update that also fixed several
> >very annoying bugs any further.
>
> You've just illustrated the bodhi process working AS IT IS SUPPOSED TO. You
> had testers giving karma, and they all had positive feedback, which means that
> THE PACKAGE WAS TESTED BEFORE IT WENT TO STABLE.

Imho it is more a perversion of how it is meant to be. This package was
tested before it went to updates-testing and therefore went straight to
stable. But the majority of packages goes to updates-testing and is not
tested by someone else but the maintainer/does not get any karma, but
still is pushed to stable after some time.

Regards
Till
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 12:55 PM
Kevin Kofler
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

Michael Schwendt wrote:
> That would be a ridiculous decision. It would be much better to disable
> that feature only for those update submitters who really have been
> dilettantish enough to use it inappropriately more than once.

Yeah, that's a good idea. We really need to avoid punishing everyone for the
few incompetent maintainers who screw up!

>> * A new package which doesn't replace anything, and which I verified to
>> work fine for me. It's clearly not a completely broken package and
>> there's no way it can break anybody's existing setup as nobody has that
>> package yet.
>
> Unconvincing, though. History has shown that some packagers still managed
> to push new packages that suffered from broken deps and failure to start
> at run-time (and even misplaced files). Especially for _new_ packages,
> dep-breakage would be avoidable by pushing them as test-updates as long as
> there is not integrated depchecking yet.

Well, as I wrote, the packager should have tested the package he's pushing
out, of course! Especially for a new package, it's the only way to know it
works. Something that doesn't work at all has no business being pushed to
anywhere, even testing. And yes, checking that the dependencies are all in
Fedora is definitely a good idea, too. (But automated depchecks would solve
that problem once and for all.)

>> * A regression which causes big breakage at least for some people slipped
>> through testing for whatever reason. We urgently want the fix to get out
>> ASAP.
>>
>> * A regression slipped through testing for whatever reason and the patch
>> is trivial. We want the fix to get out ASAP, and the risk of breakage is
>> very low.
>
> The possibility to publish hot-fixes is most important.

+1. Not being able to push those out quickly would really suck.

>> * A trivial bugfix (like a one-line diff), tested and confirmed to fix
>> the bug by at least one person. The risk of breakage is extremely low.
>
> For some bugs and some bug-fixes [and some packages], waiting for testers
> is just a waste of time.

Indeed.

Kevin Kofler

--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 01:04 PM
Jaroslav Reznik
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

On Friday 26 February 2010 14:32:16 Marcela Maslanova wrote:
> ----- "Josh Boyer" <jwboyer@gmail.com> wrote:
> > On Fri, Feb 26, 2010 at 08:14:13AM -0500, Marcela Maslanova wrote:
> > >----- "Matthias Clasen" <mclasen@redhat.com> wrote:
> > >> On Fri, 2010-02-26 at 13:16 +0100, Kevin Kofler wrote:
> > >>
> > >> I think banning stable pushes is the right idea. None of your
> >
> > reasons
> >
> > >> is
> > >> very convincing.
> > >
> > >My packages are rarely tested and I forget them in testing phase for
> >
> > a
> >
> > >long time. Also fixing BR don't need testing. I simply need push
> > >immediately the new/fixed package.
> >
> > If nobody is testing your packages sitting in updates-testing, then
> > maybe the
> > users of that package aren't hitting whatever you're fixing or aren't
> > otherwise
> > having other issues. What is the benefit of pushing an update if
> > nobody cares?
>
> They don't care about bodhi and probably they don't know about it.

Good point - it's useless to have package sitting in updates-testing if there
are no users to test it because they usually don't know that we have Bodhi.
Maybe some package rating included in PackageKit would be nice - for stable
packages it's indicator that this package is worth to install, for testing
package it would mean it's working (but again - who's going to rate it in
pkgkit once installed).

I think it's still worth to go through updates-testing with exceptions (eg 95%
testing). And yes - lot of issues could be solved with auto QA.

At least I try to test packages in updates-testing on my other systems
(running different Fedora versions). And thanks to scripts finding my f*kups
like broken deps.

Jaroslav
--
Jaroslav Řezn*k <jreznik@redhat.com>
Software Engineer - Base Operating Systems Brno

Office: +420 532 294 275
Mobile: +420 731 455 332
Red Hat, Inc. http://cz.redhat.com/
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 01:04 PM
Chris Adams
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

Once upon a time, Kevin Kofler <kevin.kofler@chello.at> said:
> at the FESCo meeting on Tuesday, everyone except me seemed to be set on

Do you really see _everything_ as FESCo (or the world) vs. Kevin Kofler?
I read over the FESCo logs from time to time, and your repeated
foot-stomping on the DSO linking change was rather childish, and now we
have this.

> (We really need more transparency in decision
> making!)

So to you, stirring up controversy without an actual proposal is
"transparency"? How much more transparent can open and logged meetings
get?

You clearly want to be able to push whatever, whenever (see massive KDE
updates in supposedly "stable" releases). Others have shown that
playing fast and loose with updates has consequences, and putting a
little delay in there would probably be better.

EPEL has run this way for a while, and it doesn't seem to be a problem.
--
Chris Adams <cmadams@hiwaay.net>
Systems and Network Administrator - HiWAAY Internet Services
I don't speak for anybody but myself - that's enough trouble.
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 01:28 PM
Orcan Ogetbil
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

On Fri, Feb 26, 2010 at 9:24 AM, Michael Schwendt wrote:
> On Fri, 26 Feb 2010 08:26:59 -0500, Orcan wrote:
>
>> Another annoying issue is updates with no explanations. There is a
>> "Notes" field in bodhi that many people just ignore for an unknown
>> reason. Any update with less than a specified number of characters
>> (~40) in the Notes should also be banned.
>
> Nonsense. Such arbitrary rules will only drive off packagers. The field in
> an update request may be empty because the list of bugzilla tickets is
> sufficient and because the package %changelog adds further details.
>

If those were sufficient, blank Notes wouldn't be annoying, right? But
it is annoying. Hence my proposal.

Orcan
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 01:39 PM
Chris Adams
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

Once upon a time, Patrice Dumas <pertusus@free.fr> said:
> On Fri, Feb 26, 2010 at 08:04:55AM -0600, Chris Adams wrote:
> > EPEL has run this way for a while, and it doesn't seem to be a problem.
>
> EPEL is very different. Packages in EPEL have been tested in fedora and so
> will very rarely need hotfixes aor regression fixes (except for security
> fixes, which if I recall well are covered by an exception already).

That reasoning would tend to argue for the reverse of EPEL's policy. If
the packages are so stable, why go through testing?

Every time a package is built, it is susceptible to new bugs. Packaging
bugs, build requirement changes, and software bugs all creep in, and not
trying to ram things out the door as fast as possible seems like a good
idea.
--
Chris Adams <cmadams@hiwaay.net>
Systems and Network Administrator - HiWAAY Internet Services
I don't speak for anybody but myself - that's enough trouble.
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 01:41 PM
Tom Lane
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

Kevin Kofler <kevin.kofler@chello.at> writes:
> Some situations where I and others have used direct stable pushes in the
> past and where I think they're really warranted and should be used:

You forgot security fixes. The proposed policy is insane.

regards, tom lane
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 01:47 PM
Kevin Kofler
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

Matthias Clasen wrote:
> But presumably we still want to test the fix, to avoid introducing yet
> another regression ?!
[snip]
> Just go up to your first argument: the breage slips through. That is
> exactly what happens if your judgement of 'low risk' turns out to be
> wrong. And it will...
[snip]
> Again: go up. Breakage always happens to somebody else. That one person
> tested the fix is not enough.

Quite the opposite, the regression happened DESPITE going through testing,
it shows that testing didn't help. So why would forcing everything to go
through testing prevent breakage?

Of course evaluating the risk vs. the urgency is important, but I think
there are quite a few cases where getting the update out ASAP is the best
solution.

For example, if an application is completely broken due to an update and
nobody noticed that during testing, then why should the fix have to go
through testing? It can't make things worse if the app is already broken,
and clearly nobody is using both updates-testing and that app.

Kevin Kofler

--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 01:49 PM
Kevin Kofler
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

Josh Boyer wrote:
> If nobody is testing your packages sitting in updates-testing, then maybe
> the users of that package

… who are using updates-testing …

> aren't hitting whatever you're fixing or aren't otherwise having other
> issues. What is the benefit of pushing an update if nobody cares?

Because the people who don't run updates-testing care and complained about
the issue? Because you don't know how many more users are having the issue
and not bothering to report it (and of course if they don't even bother
reporting the issue, they don't use testing either)?

Kevin Kofler

--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
 
Old 02-26-2010, 01:59 PM
Kevin Kofler
 
Default FESCo wants to ban direct stable pushes in Bodhi (urgent call for feedback)

Orcan Ogetbil wrote:
> Another annoying issue is updates with no explanations. There is a
> "Notes" field in bodhi that many people just ignore for an unknown
> reason. Any update with less than a specified number of characters
> (~40) in the Notes should also be banned.

That's a completely unrelated issue!

I actually sorta agree with you on this point (though enforcing 40 chars
minimum is unlikely to help, it'll just lead to folks filling in crap like
"update foo to the latest upstream version", we're already seeing useless
"information" like that), but this has absolutely nothing to do with direct
stable pushes.

> I can't see a reason to make exceptions.

What about the many valid reasons that have been brought up? E.g. if a
package is destroying people's hardware, wouldn't you want the fix to go out
BEFORE your hardware is dead?

> If people used the testing repo appropriately, things would actually get
> tested.

I don't think it's the maintainers' fault that testing is insufficient, but
rather that there are just not enough testers.

> I wish there was a solution without some sort of banning, but apparently,
> there is not.

Policies? Banning only those folks who don't follow policies from direct
stable pushes (or even from Bodhi entirely, or even all of Fedora)? Why
punish those who work really hard to make things work for everyone and who
will now have their perfectly fine and safe fixes delayed for purely
bureaucratic reasons?

> Any change needs testing. Even one liners.

I'm not convinced. Some changes are really trivial.

And direct stable pushes are usually tested by at least one person before
they're queued directly to stable anyway. For a one-line fix, that's usually
more than enough (and when it's not, the maintainer knows why, e.g. if that
one line enabled a 10000-line feature).

Kevin Kofler

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

Thread Tools




All times are GMT. The time now is 09:26 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org