[cairo] 1.10 release schedule

Benjamin Otte otte at gnome.org
Thu Jan 7 01:56:45 PST 2010


Hey,

I've been wondering about the 1.10 release. I'm pretty sure we'll not
make all the ambitious goals from the hackfest[1] unless some sort of
wonder happens. (That's almost obvious after all the stuff going on
with the developers' real lifes...). But:

- I had expected at least one prerelease.
This seems to be the biggest issue: There's noone pushing for releases
or doing any sort of QA (ensuring make check keeps working, revieweing
APIs, ...). Most of the developers seem have wandered off either to
their own branches (I know at least Chris, Mozilla and me are guilty
of that) or are doing something else entirely (probably the rest of
us). I suspect we need a release manager and don't have one. I'd like
to avoid having to be that person because I do way too much other
stuff and don't know the code well enough, but if noone steps up to
doing it, I'd volunteer.

- We don't have a release schedule
I fear that if we do not commit to some sort of schedule, we'll not
get a 1.10 release out in time for the next round of distro releases.
And I really really want to avoid distros shipping 1.5 years old code
when we could do so much better. Not because of the new features or
performance, but because there's actually developers that care about
the code (see also next point).
Pixman seems to manage the 6monthly releases synced to distros pretty
well and it'd be nice if we could manage to do that, too.

- There's no branch everyone agrees on should be 1.10
In fact, we have 3 branches that could qualify: 1.8, master and
ickle/wip/compositor[2]. (Almost?) everyone agrees that 1.8 is
outdated. I know because whenever I mention a bug or performance
problem the answer is "Oh, that code. I replaced that with
$NEW_AND_AWESOME in $BRANCH". But it's the only branch that succeeds
in running the testsuite.
Then there's master. Master is quite a bit better than 1.8 (both in
perf and features), but it fails the testsuite rather spectacularly
and doesn't see much work (the last month only saw 4 tiny fixes). So
there's noone really pushing it forward.
And then there's Chris' compositor branch that is seeing active
development. It's even better than master (in performance, features,
and probably the testsuite, too). Unfortunately it's a 50.000 line
diff to master in src/ only and has no convincing git log.

So what now? Thoughts? Opinions?

Cheers,
Benjamin

PS: I actually wanted to write this with my new @redhat.com email
address to a) make this sound more official and distro-concerned and
b) include the information of my new employer for everyone who doesn't
read my blog, but I had forgotten that big companies fail in setting
up new accounts quickly.


[1]: http://gstreamer.freedesktop.org/wiki/VideoHackfest/Conclusions -
see Timeline section
[2]: http://cgit.freedesktop.org/~ickle/cairo/log/?h=wip/compositor


More information about the cairo mailing list