Hi,

Am Samstag, den 12.07.2008, 14:01 +0100 schrieb Ian Lynagh:
> On Sat, Jul 12, 2008 at 02:42:28PM +0200, Joachim Breitner wrote:
> > 
> > Theoretically, there are "Dep-Waits", i.e. flags that tell the buildds
> > that a particular package should not be tried until the other version is
> > present. These are set when a package is uploaded, and would fix our
> > problem.
> > 
> > Unfortunately, they do not work well with == build dependencies, and it
> > seems for a reason[1]
> > 
> > [1] http://lists.debian.org/debian-release/2008/07/msg00169.html
> 
> If you depend on foo == 2 and foo 3 appears in the archive, then foo 2
> never can. So if there is a build-dep on foo == 2 then dep-waiting on
> foo >= 2 would probably make most sense (for a small change). There will
> be a rebuild that is guaranteed to fail when foo 3 is uploaded, and a
> subsequent upload depending on foo == 3 will fix it without any manual
> intervention needed from the buildd folk.
> 
> Ideally, of course, you would add a dep-wait on foo == 2 and turn that
> into failed if foo 3 is uploaded, but that is presumably a larger change
> needed to the autobuilder software.

Actually, that can be implemented easily, it seems:
http://lists.debian.org/debian-release/2008/07/msg00260.html

But it seems to me that we do not gain anything from using == instead of
>=. OTOH, if we would use >=, we could make haskell-hgl now installable
by either doing a new sourcful upload, or by requesting binNMUs for all
arches – that adds flexibility.

In any case, the current system (== with no change to wanna-build) is
not sustainable, at least not if you ask one of the buildd admins: :-(
http://buildd.debian.org/~jeroen/status/package.php?p=xmonad-contrib#fail-xmonad-contrib-s390

Greetings,
Joachim
-- 
Joachim "nomeata" Breitner
Debian Developer
  [EMAIL PROTECTED] | ICQ# 74513189 | GPG-Keyid: 4743206C
  JID: [EMAIL PROTECTED] | http://people.debian.org/~nomeata

Attachment: signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil

_______________________________________________
debian-haskell mailing list
[email protected]
http://urchin.earth.li/mailman/listinfo/debian-haskell

Reply via email to