[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#354674: What on earth?



In-Reply-To: <20060416145221.GB7942@localdomain>
On Sun, Apr 16, 2006 at 10:52:21AM -0400, David Nusinow wrote:
> As for the build-depends, pbuilder is, as far as I've been able to
> understand it, completely incapable of handling such a massive beast as
> this. You can't point it easily at a custom repository in order to have it
> pull from there. If this has changed recently, I'd love to hear it, but
> when I was investigating this during the development of the packages I was
> unable to do it. Furthermore, the packages I pulled were autobuilding just
> fine on Ubuntu, so I had little reason to believe that they didn't have
> proper build-depends for Debian. Indeed, very few of the packages ftbfs'ed,
> and most of these were fixed within hours of being reported.

  AFAICT, it's easy to achieve. See [1]. bindmount your
/var/cache/pbuilder/results (or the place where you put your built
packages) e.g. in /mnt/results/. Also add a hookscript starting with D
that looks like:

  #!/bin/sh
  
  dpkg-scanpackages /mnt/results /dev/null > /mnt/results/Packages
  echo "file:///mnt/results ./" >> /etc/apt/sources.list
  apt-get update

and I think you are done. I reckon this was not in pbuilder(8) but in
the online user doc.

> > I can predict that the Xorg 7.0 will be the messiest debian will have to 
> > face in years, because everything is done in a hurry, and that each new 
> > uploads adds as many bugs (if not twice as many) as it solves.
> 
> I'd like for you to back this claim up. So far I've fixed dozens of bugs
> over the course of the past week at great personal and professional cost of
> my time, energy, and health. And I plan to keep whittling away at the bugs
> until the transition is as clean as I can possibly make it.

  well, like said, it's a bit late for that. And about your bug load,
let me say it loud clear: the current load you are experiencing was
created by your way to package X.org 7.0, not anticipating any of the
problems.  I won't say your current rate of bug fixing isn't remarkable
outside from the current contexte. But please, when someone run into a
wall, nobody will think that his survival is a remarkable thing.

> > Could please the XSF communicate, and announce what that damn transition 
> > implies for *everybody*, instead of letting anybody finds out that 
> > their package is broken. I suggest [1] as a very good template for what 
> > communicating about a transition means.
> 
> I communicated, to the best of my knowledge, what the transition meant in
> the past [0]. I wasn't aware that I would be breaking a large amount of
> packages until after I uploaded to unstable. These packages have largely
> been in use in Ubuntu for several months already. They have been in
> experimental for several months as well, during which time I fixed every
> bug that came in about them. I communicated with the release team what my
> plans were at all stages, and while I didn't realize the scope of the
> disruption, I did my absolute best to keep everyone involved informed.
> Every single change I've done to these packages has been documented on
> debian-x via the svn commits, so everyone could see what I've been doing if
> they cared to look.

  I'm shocked that you didn't anticipate *any* of the problems you ran
into. after all, you've broken : *dm, fonts, made FTBFS a lot of
programs linking against xlibs, ...  That looks to me like beeing an
reasonnably complete coverage of the things X.org is supposed to
achieve. The affirmation "This worked fine in ubuntu" looks like a very
loose quality quality test for a first upload of a totally new layout of
an X server.

> I have yet to file bugs against other packages that I've affected because I
> haven't had time. [...]

  that should have been done *before* any of your moves. And that's
exactly the RM's point, and mine also.

> Anyway, I'm going to continue to work hard on this. If you want to help dig
> us out of it, I'll welcome any patches you care to submit that are up to
> your standards of quality.

  I'm already fixing a huge load of RC bugs doing NMU to clean packages
that are in a loosy shape since a lot of time, and that does not seem to
have an active enough maintainer.  I hope X.org does not qualify yet to
those criteriums.

> Until then though, feel free to kiss my ass.

  You're welcome.

  [1] http://www.netfort.gr.jp/~dancer/software/pbuilder-doc/pbuilder-doc.html#id267034



In-Reply-To: <20060416134720.GH18401@fooishbar.org>
> >  - /usr/X11R6/bin/X disapearing broke login managers (gdm, kdm)
>
>  This is being rectified, as a perfunctory glance at -x will tell you.

  Are you serious ? Do you read debian-glibc@ each time you upload a
package ? please, if something has te be known related to a migration,
it has to come from the team that launched that migration.  It's not to
the others developpers to go read -x, it's up to you to inform them.

  So I really expect that in a really near future I'll hear from you (as
a team member of the QT-KDE team that packages kdm), that gdm maintainer
will hear from you as well, and that font packagers will see bugs go
through the BTS to ask them to change their paths.

> > So maybe it's now time to calm down the upload rate (yeah unstable is 
> > broken, but it's too late for that anyway, and after all it's not 
> > called unstable for nothing), let's have some communication to have it 
> > fixed, instead of pile of clumsy patches.
> 
> So, let me get this straight: on one hand you're complaining about bugs,
> and on the other hand, you're complaining about bugs being fixed?  The
> workload of the XSF members getting things fixed is very admirable.

  I'm complaining because *you* created the huge load of bugs you have
to cope with, and a lot of other you don't warn other packagers about
(what pissed me, and made me write my previous mail is yet-another-RC
bug because of X we received on kdm recently...).  And also in your
answer to me, you ask to be sanctified because you are currently in a
hurry to fix them ?

  heh, something does not looks right to me.

> David has posted a couple of messages on debian-devel-announce
> discussing the transition (including xlibs-dev), and what it means for
> everyone.  Most of the transition was co-ordinated in excruciating
> detail, including a long time in experimental where testing failed to
> uncover these sorts of problems.

  I guess we do not read the same list, because the last mail from David
concerns the xlibs-dev transition for Xorg6.9[2]. I've not seen anything
related to Xorg modular transition yet, and that's why I'm complaining,
because I'm in the dark here, and that instead of knowing why I suffer
packaging things against Xorg, I just discover problems I should have
been warned against in the first place.

> (Hence the delay in experimental, waiting for testers.)

  Give me a break. For one single package, it's already quite penible to
use experimental (to avoid to pull every single experimental package,
you have to edit your /etc/apt/preferences, and stuff like that), it's
not imagineable that users will use experimental for that reason,
because with the myriad of libraries that X comes with, either you pull
all experimental (who is insane enough to only think of doing that ?) or
edit a 645-line long /etc/apt/preferences, which nobody will want to do
either.

  I honnestly think things have been rushed[3] too much, as many of the
problems that have been raised, could have been found, warned, or worked
around in early packaging stages (meaning before an upload).


  [2] http://lists.debian.org/debian-devel-announce/2006/01/msg00003.html
  [3] http://lists.debian.org/debian-project/2004/01/msg00099.html

-- 
·O·  Pierre Habouzit
··O                                                madcoder@debian.org
OOO                                                http://www.madism.org

Attachment: signature.asc
Description: Digital signature


Reply to: