Bug#699808: tech-ctte: syslinux vs the wheezy release
On Tue, Feb 05, 2013 at 01:48:22PM +0100, Julien Cristau wrote:
> the background for this request can be found in bug#699382. Here are
> the highlights:
> - the debian-installer source package, which builds the installer images
> for debian's releases, build-depends on syslinux
> - the release freeze for wheezy started in June 2012, and is now in its
> final stages
> - one of the prerequisites for the release is a release candidate for
> the installer
> - the syslinux maintainer uploaded new upstream versions of his package
> to unstable, which were unsuitable for wheezy, in November 2012, and
> again at the end of January 2013
> - the latest of these uploads breaks the installer, making it impossible
> to build and upload the planned wheezy release candidate, since
> build-dependencies are fetched from unstable
> - when asked to revert this change, the syslinux maintainer refused, and
> said disagreements should be referred to the technical committee
>
> The submitters consider this disregard for the Debian release process
> not acceptable, and ask that the technical committee
> - overrule the syslinux maintainer, and decide that the syslinux package
> in unstable should be reverted to the version currently in wheezy (or a
> version compatible with the release team's freeze policy), until
> the wheezy release
> - rule that the syslinux package maintainer should be willing to
> collaborate with maintainers of its reverse-dependencies (and in
> particular the Debian Installer team) in a way that's compatible with
> Debian's release processes, and not against them.
Even if the process is broken (and I can see arguments either way on
that), now is not the time to rearrange it. If it was known in advance
that d-i et al would break, at the very least patches for that should
have been landed first. If it was not known, then what makes us think
that the three problems identified to date (d-i, debian-cd, virtualbox)
are a complete list? At the very least, it puts into question a
considerable amount of the installer testing done to date.
I would not wish to "rule that the syslinux package maintainer should be
willing to collaborate" etc.; this is the type of assuming-bad-faith
element of rulings that I think has not generally gone well when the TC
has attempted it in the past, and it's at best questionable whether it's
within our power anyway. But I do think that the syslinux maintainer
should revert to 4.x in unstable; I'd rather that be voluntary but I'd
be willing to vote to overrule if need be.
As for the reasoning Daniel gave for putting 5.x in unstable in the
first place, my recommendation would be to put 5.x in experimental and
6.x in some other archive; they are hardly difficult to set up and I'm
quite sure Daniel's already familiar with what's involved there. The
additional marketing gained by having 6.x in the Debian archive proper
doesn't justify indirectly causing problems for wheezy, even if one
disagrees with the process by which those problems are caused.
On a personal note, this is exactly the kind of reason I deliberately
kept GRUB 1.99 in unstable and 2.00 in experimental, even though it
would have been more convenient and rather more pleasing to be able to
ship wheezy with 2.00. There are certainly cases where new upstream
versions in unstable during a freeze don't cause a practical problem and
needn't involve any fuss - I've uploaded some such myself - but once
it's clear they do cause a practical problem the proper course of action
is surely to revert and let everyone get on with pre-existing problems.
Cheers,
--
Colin Watson [cjwatson@debian.org]
Reply to: