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 01-05-2010, 04:38 PM
Jerry James
 
Default Our static Libraries packaging guidelines once more

On Tue, Jan 5, 2010 at 10:23 AM, Daniel P. Berrange <berrange@redhat.com> wrote:
> That sounds good as long as AutoQA is reliable, not generating false
> positives. I'd still also suggest that we have a rule drop all
> packages reported by the FTBFS tests which aren't fixed by time of
> Beta.

What about packages that fail to build because they depend on some
other package that is broken? I've got one in that state now. It
fails to build from source because one of its BuildRequires is broken.
There's nothing wrong with my package. Once the other guy fixes his,
mine will magically start building again. If the other guy hasn't
fixed his package by Beta, how is dropping mine going to help?
--
Jerry James
http://www.jamezone.org/

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-05-2010, 04:49 PM
"Daniel P. Berrange"
 
Default Our static Libraries packaging guidelines once more

On Tue, Jan 05, 2010 at 10:38:50AM -0700, Jerry James wrote:
> On Tue, Jan 5, 2010 at 10:23 AM, Daniel P. Berrange <berrange@redhat.com> wrote:
> > That sounds good as long as AutoQA is reliable, not generating false
> > positives. I'd still also suggest that we have a rule drop all
> > packages reported by the FTBFS tests which aren't fixed by time of
> > Beta.
>
> What about packages that fail to build because they depend on some
> other package that is broken? I've got one in that state now. It
> fails to build from source because one of its BuildRequires is broken.
> There's nothing wrong with my package. Once the other guy fixes his,
> mine will magically start building again. If the other guy hasn't
> fixed his package by Beta, how is dropping mine going to help?

It will motivate you, or someone else depending on it, to become a
co-maintainer of the broken package & help with fixing it ;-P In
all seriousness though, it is very bad if we're having many of cases
of large sets of downstream package chains being blocked by an dependant
one failing. If a security issue arises in the FTBFS package we're
between a rock & a hard place, which is why I think it is worth being
strict on fixing FTBFS bugs.

Regards,
Daniel
--
|: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :|
|: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :|
|: http://autobuild.org -o- http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-05-2010, 05:07 PM
Jesse Keating
 
Default Our static Libraries packaging guidelines once more

On Tue, 2010-01-05 at 12:27 -0500, Tom Lane wrote:
> "Tom "spot" Callaway" <tcallawa@redhat.com> writes:
> > On 01/05/2010 11:30 AM, Jesse Keating wrote:
> >> On Tue, 2010-01-05 at 11:03 -0500, Tom Lane wrote:
> >>> On the other hand, with the
> >>> guideline being so widely ignored, I'm not in a hurry to do work to
> >>> comply with it ...
> >>
> >> Isn't that a chicken/egg problem?
>
> > It really is.
>
> Well, fwiw, I have to fix the same two spec files for the %define
> problem, so I'm going to take care of this today while it's fresh in
> mind. But there's a general issue that new things keep getting added
> to the packaging guidelines and there's no very good mechanism to
> detect whether existing packages ever get updated to comply.
>
> regards, tom lane
>

In the future when we have AutoQA online we'll be able to add new tests
for new guidelines and alert maintainers who's specs fall out of.. er..
spec.

--
Jesse Keating
Fedora -- Freedom˛ is a feature!
identi.ca: http://identi.ca/jkeating
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-05-2010, 05:13 PM
Ralf Corsepius
 
Default Our static Libraries packaging guidelines once more

On 01/05/2010 05:48 PM, Tom "spot" Callaway wrote:

On 01/05/2010 11:30 AM, Jesse Keating wrote:

On Tue, 2010-01-05 at 11:03 -0500, Tom Lane wrote:

On the other hand, with the
guideline being so widely ignored, I'm not in a hurry to do work to
comply with it ...


Isn't that a chicken/egg problem?


It really is. I mean, we could create the "Packaging Police" to run
around and enforce the guidelines by force (either by fixing them
manually,
I would not want to call it a "packaging police", but a "tag team" which
fixes the packages


== IMO, that's the way to go.


or by threatening maintainers until they do it), but is that
really what we want?


Yes, people have had enough time to fix their packages - It's time for
action.


Ralf

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-05-2010, 08:20 PM
Till Maas
 
Default Our static Libraries packaging guidelines once more

On Tue, Jan 05, 2010 at 11:48:47AM -0500, Tom spot Callaway wrote:
> On 01/05/2010 11:30 AM, Jesse Keating wrote:
> > On Tue, 2010-01-05 at 11:03 -0500, Tom Lane wrote:
> >> On the other hand, with the
> >> guideline being so widely ignored, I'm not in a hurry to do work to
> >> comply with it ...
> >
> > Isn't that a chicken/egg problem?
>
> It really is. I mean, we could create the "Packaging Police" to run
> around and enforce the guidelines by force (either by fixing them
> manually, or by threatening maintainers until they do it), but is that
> really what we want?

I would skip the threatening part, but allowing provenpackagers to fix
violations to the packaging guidelines after a short notice to the
maintainers is something we should encourage imho. It just plain sucks
if there are bugs that can be fixed easily and may cause issues, but
it takes several weeks to be able to fix them oneself.



Regards
Till
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-05-2010, 09:57 PM
Till Maas
 
Default Our static Libraries packaging guidelines once more

On Tue, Jan 05, 2010 at 05:23:08PM +0000, Daniel P. Berrange wrote:

> That sounds good as long as AutoQA is reliable, not generating false
> positives. I'd still also suggest that we have a rule drop all
> packages reported by the FTBFS tests which aren't fixed by time of
> Beta.

I would like to have this with a slight modificiation. If a package
FTBFS for at least a certain amount of time (e.g. two weeks) at the time
of Beta, then every provenpackager may just fix the bugs for another
certain amount of time (e.g. another two weeks) and if nobody fixes it
then it should be dropped. Or maybe we could have some kind of
"neglected packages task force", that may just in general fix bugs in
packages that are not fixed by the original maintainer. The advantage
over becoming co-maintainer of certain packages is then, that one does
not get all the noise about bugs that are already been taken care of by
the original maintainer.

Regards
Till
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-06-2010, 12:57 PM
Adam Williamson
 
Default Our static Libraries packaging guidelines once more

On Tue, 2010-01-05 at 12:16 -0500, Tom "spot" Callaway wrote:

> Well, I think a reasonable alternative would be to add those policies to
> the AutoQA infrastructure, and if the package fails the check, it
> doesn't get tagged and the packager gets an email explaining the
> failure. That will get things fixed up.

The only problem with that is that just about every packaging guideline
has _some_ valid exceptions (that's why they're all guidelines...) and
it's rather hard to build exceptions into an automatic testing system in
a way which doesn't get horribly crufty in a hurry. But yes, broadly I'm
in favour of this kind of thing. Mandriva does it to a limited extent (a
few rpmlint checks are run on submitted packages and certain failures
cause the package to be rejected) and it does stop people making really
bad mistakes.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Fedora Talk: adamwill AT fedoraproject DOT org
http://www.happyassassin.net

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-06-2010, 03:38 PM
Michael Schwendt
 
Default Our static Libraries packaging guidelines once more

On Wed, 06 Jan 2010 13:57:14 +0000, Adam wrote:

> On Tue, 2010-01-05 at 12:16 -0500, Tom "spot" Callaway wrote:
>
> > Well, I think a reasonable alternative would be to add those policies to
> > the AutoQA infrastructure, and if the package fails the check, it
> > doesn't get tagged and the packager gets an email explaining the
> > failure. That will get things fixed up.
>
> The only problem with that is that just about every packaging guideline
> has _some_ valid exceptions (that's why they're all guidelines...) and
> it's rather hard to build exceptions into an automatic testing system in
> a way which doesn't get horribly crufty in a hurry.

If exceptions become a problem because they are applied to many packages,
it would still be possible to adjust the guidelines or mark the packages
with special metadata comments in their .spec files. Then packagers would
need to make use of an exception _explicitly_, showing that what they do
is intentional.

One example where the packager was mistaken about an exception in the
guidelines is "mpich2":
https://bugzilla.redhat.com/buglist.cgi?quicksearch=Violation+of+the+Packaging +Guidelines

From the few tickets I had filed in early December (for early corner-cases),
only 60% have been fixed. No sign of activity in the remaining tickets.

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-06-2010, 04:02 PM
"Daniel P. Berrange"
 
Default Our static Libraries packaging guidelines once more

On Wed, Jan 06, 2010 at 01:57:14PM +0000, Adam Williamson wrote:
> On Tue, 2010-01-05 at 12:16 -0500, Tom "spot" Callaway wrote:
>
> > Well, I think a reasonable alternative would be to add those policies to
> > the AutoQA infrastructure, and if the package fails the check, it
> > doesn't get tagged and the packager gets an email explaining the
> > failure. That will get things fixed up.
>
> The only problem with that is that just about every packaging guideline
> has _some_ valid exceptions (that's why they're all guidelines...) and
> it's rather hard to build exceptions into an automatic testing system in
> a way which doesn't get horribly crufty in a hurry. But yes, broadly I'm
> in favour of this kind of thing. Mandriva does it to a limited extent (a
> few rpmlint checks are run on submitted packages and certain failures
> cause the package to be rejected) and it does stop people making really
> bad mistakes.

At time of the initial package review, the packager has to justify the
exception to the reviewer. Post-package review packager can do whatever
they want. The lack of ongoing analysis of packaging changes post-review
is a hole in our process.

If we decided to turn a certain subset of the guidelines into hard rules,
then we'd want a way to record per-package exceptions in AutoQA, along
with a short justification text. This tracking would ensure we know about
changes/issues that arise post-review, closing that hole in our process.

Regards,
Daniel
--
|: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :|
|: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :|
|: http://autobuild.org -o- http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 01-07-2010, 01:57 PM
Adam Williamson
 
Default Our static Libraries packaging guidelines once more

On Wed, 2010-01-06 at 17:38 +0100, Michael Schwendt wrote:

> > The only problem with that is that just about every packaging guideline
> > has _some_ valid exceptions (that's why they're all guidelines...) and
> > it's rather hard to build exceptions into an automatic testing system in
> > a way which doesn't get horribly crufty in a hurry.
>
> If exceptions become a problem because they are applied to many packages,
> it would still be possible to adjust the guidelines or mark the packages
> with special metadata comments in their .spec files. Then packagers would
> need to make use of an exception _explicitly_, showing that what they do
> is intentional.

Yup, indeed - this is the approach MDV uses for its rpmlint checks (you
can code an exception into the spec file if it's justified).
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Fedora Talk: adamwill AT fedoraproject DOT org
http://www.happyassassin.net

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 

Thread Tools




All times are GMT. The time now is 09:43 AM.

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