|| ||Colin Watson <cjwatson-AT-debian.org>|
|| ||Release update|
|| ||Sat, 21 Feb 2004 22:04:23 +0000|
It's been a little while since the last update on sarge. So, where are
we? Since 1 December, quite a bit of progress has been made:
* debian.org compromise recovery finished, various upgrades
The archive's open again, and with the odd exception services are
back to pre-compromise levels.
* debian-installer improving
d-i beta2 came out of the door in mid-January, with support for
i386, ia64, and powerpc. alpha and mips are getting there, with some
reports of successful installations by developers. hppa, m68k, and
mipsel are improving but still need work; s390 and sparc are behind,
although there's been recent activity; arm seems to be further
behind still. (My apologies if these details are out of date, and
please correct me.)
I believe we're doing pretty well here. We probably need another
beta iteration or two, but the point where we could release with it
seems to be in sight.
* GNOME 2.4 in testing
After a lot of work, the GNOME 2 metapackages are finally
installable in testing.
* KDE 3.1.5 in testing
Over the last few weeks, KDE 3.1.5 has been approaching the state
where it's ready for testing. Yesterday Anthony overrode the last
few problems with the underlying jack-audio-connection-kit
dependency chain, so the KDE metapackages are also finally
installable in testing. Only a few components remain at older
* XFree86 4.3 in unstable
A number of people chipped in to finish XFree86's patch audit, which
is now done, and 4.3.0-2 was uploaded to unstable on Wednesday.
On the other hand, things that still need to be done:
* Finish debian-installer and port it to remaining architectures
If you want to help sarge release and you're interested in anything
on Joey's task list (including testing), go and help d-i now.
As Joey reported, a number of ports are still lagging. Developers on
those ports should go and work on getting themselves a working
installer as soon as possible: ultimately, we may just have to
decide to release without architectures that lag too far behind on
* Remaining tweaks to desktop environments
We've finally got completely installable desktop environments in
testing. Now's the time for final adjustments and (minor!) upgrades.
* Finish XFree86
I imagine we should expect a few final cleanups and bug fixes before
XFree86 4.3.0 can be released with sarge. Let's budget for that,
* Finish miscellaneous library changes
There are a few library soname changes that have yet to make it into
testing, and need to do so.
* Release-critical bugs
We've got 350-odd release-critical bugs on packages in testing at
the moment, and this number has if anything been increasing for the
last month or so. This isn't something we can release with.
So, much like in Anthony's 1 December post, we're not in terribly bad
shape, but we have a lot to do.
It's instructive to have a look at the RC bugs graph once again, if you
haven't done so recently:
You'll see that the end of 0-day NMUs on 12 January pretty much
coincides with a plateau in the downward slope of the bug count. It
seems clear that Anthony's comment in December still applies:
Without having evaluated null hypotheses or done exhaustive analyses,
the correlation nevertheless seems fairly convincing. To put it
bluntly, our regular package maintainers are doing such a bad job that
without significant assistance from NMUs, about 6% of the archive
fails to meet even our _absolute minimum_ expectations.
Please check that you have no open release-critical bugs (that aren't
tagged in some way indicating that they don't affect sarge). If you do,
fix them. If you can't fix them, ask for help. If you've already asked
for help and nobody's responded, consider whether the package really
needs to be in sarge. If it really does need to be in sarge, shout
louder for help.
At this point, it is simply not good enough for critical, grave, or
serious bugs to be open (without sarge-ignore tags or similar) for
significant periods of time. You should expect packages with such bugs
to start being removed soon, where possible; indeed, some removals have
already been happening. I recommend that maintainers use the
'grep-excuses' tool from devscripts on a regular basis to see what's
happening with their packages in testing.
So, here's an updated schedule, intentionally without looking too far
into the future:
* now - 15th March:
Avoid library soname changes and similar.
Avoid major changes to the base system (anything installed by
Final translation changes to base system.
All release architectures should report successful d-i installs, and
be basically solid. At least i386 needs to be as bulletproof as
Drop number of release-critical bugs to 200 (by fixes and removals).
* 15th March - ???:
Last changes (without regressing translations) to packages in the
base system uploaded to unstable.
Beta testing of installation. Last-minute fixes and changes to d-i.
XFree86 4.3 should have reached testing by now.
Drop number of release-critical bugs to 100 (by fixes, removals, and
On or shortly after 15th March, we'll see if these targets have been met
and update the schedule accordingly. After that, we're looking at
freezing the rest of the system, starting test cycles, doing final
installer fixes, and sorting out security updates for sarge. I'm
deliberately not trying to set a release date here, the idea being that
it's better to wait until we're far enough along to be reasonably sure
that the date is achievable; but if you want to apply bug fixes in time
for sarge then please do so now rather than later, with care.
If in doubt about what you should or shouldn't be changing, please
Colin Watson [firstname.lastname@example.org]
Debian Release Assistant, on behalf of the release team
to post comments)