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 06-29-2008, 08:38 AM
Matej Cepl
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

https://admin.fedoraproject.org/updates/F9/FEDORA-2008-5819

Why in the world the package was pushed to stable, when it is
known to be broken? (and for the record, I don't think Marcela
did it -- she is usually too smart to work on Saturdays ;-)).

Matěj

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 06-29-2008, 10:11 AM
"Thomas Moschny"
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

2008/6/29 Matej Cepl <mcepl@redhat.com>:
> https://admin.fedoraproject.org/updates/F9/FEDORA-2008-5819
>
> Why in the world the package was pushed to stable, when it is
> known to be broken?

Funny enough, on x86_64 yum decides to install tcl-1:8.5.1-4.fc9.i386
in addition to updating to tcl-1:8.5.2-2.fc9.x86_64, in order to
solve the dependency problem, which seems to be the wrong solution to
me.

Known multilib problem?

- Thomas

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 06-29-2008, 12:22 PM
Kevin Kofler
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

Matej Cepl <mcepl <at> redhat.com> writes:
> Is really bodhi so stupid^H^H^H^Hnon-smart? [from the subject]

Yes, you have to follow procedures, which unfortunately Marcela didn't. Can
somebody in the Brno office please explain her how dist-f9-override works? :-)

> Why in the world the package was pushed to stable, when it is
> known to be broken?

Because it was already submitted for stable and the push didn't get canceled
when you sent your -1 (presumably because it was Saturday and Marcela didn't
see it).

> (and for the record, I don't think Marcela did it -- she is usually too smart
> to work on Saturdays ).

The update was submitted on "2008-06-26 06:26:25", that's Friday morning,
presumably the push to stable was requested right there.

Kevin Kofler

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 06-29-2008, 06:30 PM
Ignacio Vazquez-Abrams
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

On Sun, 2008-06-29 at 10:38 +0200, Matej Cepl wrote:
> https://admin.fedoraproject.org/updates/F9/FEDORA-2008-5819
>
> Why in the world the package was pushed to stable, when it is
> known to be broken? (and for the record, I don't think Marcela
> did it -- she is usually too smart to work on Saturdays ;-)).

https://fedorahosted.org/bodhi/ticket/79

--
Ignacio Vazquez-Abrams <ivazqueznet@gmail.com>

PLEASE don't CC me; I'm already subscribed
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 06-30-2008, 05:55 AM
Matej Cepl
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

On 2008-06-29, 12:22 GMT, Kevin Kofler wrote:
> Matej Cepl <mcepl <at> redhat.com> writes:
>> Is really bodhi so stupid^H^H^H^Hnon-smart? [from the subject]
>
> Yes, you have to follow procedures, which unfortunately Marcela didn't. Can
> somebody in the Brno office please explain her how dist-f9-override works? :-)

Could you elaborate on this, please, or point to some
explanation? I don't understand how dist-f9-override works myself
... :-(

Matěj

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 06-30-2008, 07:07 AM
Kevin Kofler
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

Matej Cepl <mcepl <at> redhat.com> writes:
> Could you elaborate on this, please, or point to some
> explanation? I don't understand how dist-f9-override works myself
> ... :-(

The problem is that updates to released distributions don't automatically end
up in the buildroots for dependent packages to build against. That's because
updates may be revoked, so usually it is safer to build against the latest
stable version in order not to introduce accidental dependencies on updates
which don't actually get pushed. Only when an update moves to the stable
updates, it ends up in the buildroot.

So until there it sounds all reasonable, but then where's the problem? Well,
sometimes libraries break compatibility both ways, meaning a build against the
old version will not run against the new one. (Usually, this happens when a
library bumps their soname, Tcl is a bit special there, but it has the same
effect.) Another potential problem is that you want to push a new version of an
application using the library together with the new library version, but that
new application version needs the new version of the library to build. (That's
the regular situation for KDE.) And of course we don't want to push the update
to the stable updates before the dependencies are being rebuilt (otherwise we
end up with the problem which started this thread)...

So what now? This is where Release Engineering (rel-eng) enters into action.
Rel-eng has the power to force packages from updates-candidate or
updates-testing into the buildroot. They do this by tagging the package in Koji
with an override tag for the distribution, i.e. currently dist-f9-override or
dist-f8-override. The buildroot will prefer the packages with the override tag
to the ones from (stable) updates (even if they're older, so it's important to
remove the override tags when no longer needed, but rel-eng normally takes care
of that). That in turn allows dependent packages to be rebuilt so they can be
pushed all at once.

Thus, the workflow is as follows:
1. build the library (DO NOT submit an update yet!)
2. send an e-mail to rel-eng asking to tag the library with dist-f9-override
3. wait for rel-eng to (manually) process the request, you'll normally get both
a confirmation mail from rel-eng and an automated tagging notification from
Koji (the mail from rel-eng goes to whomever requested the tagging, the Koji
notification goes to whomever built the library)
4. wait for Koji to complete its newRepo task (normally takes less than an
hour)
5. build the package(s) depending on the new library
6. for more complicated dependency chains, repeat steps 2 to 5 as often as
necessary
7. submit a grouped update with all the affected packages through the Koji web
interface
(If you don't have commit access to the dependent packages, you should request
it so you can do 5. and 7., otherwise some coordination with the other affected
maintainers is needed.)

Kevin Kofler

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 06-30-2008, 03:10 PM
Matej Cepl
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

On 2008-06-30, 07:07 GMT, Kevin Kofler wrote:
> The problem is that updates to released distributions don't
> automatically end up in the buildroots for dependent packages
> to build against. That's because updates may be revoked, so
> usually it is safer to build against the latest stable version
> in order not to introduce accidental dependencies on updates
> which don't actually get pushed. Only when an update moves to
> the stable updates, it ends up in the buildroot.

Thanks for the explanation, but couldn't this tck/tk
non-complicated case solved just by chain-build?

Matěj

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 06-30-2008, 03:34 PM
drago01
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

On Mon, Jun 30, 2008 at 5:10 PM, Matej Cepl <mcepl@redhat.com> wrote:
> On 2008-06-30, 07:07 GMT, Kevin Kofler wrote:
>> The problem is that updates to released distributions don't
>> automatically end up in the buildroots for dependent packages
>> to build against. That's because updates may be revoked, so
>> usually it is safer to build against the latest stable version
>> in order not to introduce accidental dependencies on updates
>> which don't actually get pushed. Only when an update moves to
>> the stable updates, it ends up in the buildroot.
>
> Thanks for the explanation, but couldn't this tck/tk
> non-complicated case solved just by chain-build?

nope chain-builds only work in devel

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 06-30-2008, 09:03 PM
Matej Cepl
 
Default Is really bodhi so stupid^H^H^H^Hnon-smart?

On 2008-06-30, 15:34 GMT, drago01 wrote:
> nope chain-builds only work in devel

Auch. OK, thanks for explanation.

Matěj

--
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 09:38 AM.

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