X.org Packaging Roadmap
I'd like to hammer out a roadmap for getting X.org packages ready to go.
Feedback from everyone is critical, of course. Here's what I've got so far:
1) Import Ubuntu packaging for X.org in to our svn repo wholesale. Whether
this should be from the Hoary release package, or the most recent
develpoment tree, I don't know, and I'd like to hear from Daniel on
what he thinks.
2) Make any necessary changes for them to be Debian, rather than Ubuntu
packages. These should not be major, but more akin to a re-branding.
3) Make sure they build on x86, ppc, and whatever else we have lying
around collectively. Also make sure they install properly.
4) Upload to experimental. Not only have there been positive reports about
updating from the Debian packages to the Ubuntu ones so far, but it's
also experimental, so we can break stuff ;-) Do one or two more uploads
to experimental as necessary to get the packages basically working the
way we want to, but in no way do we hold ourselves to shippable quality
at this stage.
5) With packages in place, we can go about making sure that all the
changes made to the Ubuntu packages are good for Debian. I have complete
faith in Daniel and Fabio's work, but I think we need to re-evaluate it. I
know Branden has expressed that he wants to be aware of what's been done,
and I need a chance to learn the X server, so this will be a good
opportunity for both.
During this phase, we can also do things like add the massive
commentary to debian/control that Branden has begun already in the
X.org branch of his svn repo. I think this is a good idea personally,
but it's not pressing that it's available in the first upload to
experimental.
6) Upload a new revision of the packages to experimental. These will have
been audited to everyone's liking, and will have whatever basic changes
are necessary. No major alterations to the packaging will have been
done.
7) Solicit builds from alternate architectures that haven't been built,
and also solicit testing from these arches. Make sure that it runs well
on all arches. One caveat to this is that I don't feel it's worth
waiting for arches that can't or won't test X when called. This is a
major package, and if lesser arches won't at least try to build the
thing within two weeks, we simply don't worry about them.
8) If the packages are, if not perfect, at least of high enough quality,
upload to unstable.
Now, I haven't been following the xorg modularization list (I did subscribe
finally though) so I don't know the progress on that. I think it'll be a
good idea to begin packaging it once things begin to take shape. Since a
lot of the work has already been done by Daniel for his debrix tree, we
should be able to figure out the basics of what we'll need to do. I think
that we should start a branch for the modular tree once the tree itself
begins to take enough of a shape that we can work on it. We can worry about
transitioning to it once we actually can see what's going on. If we release
etch with packages derived from the monolithic tree, so be it.
Comments? I'm more than happy to start the importing and such if I get svn
access, but I'd like to hear what other people think.
- David Nusinow
Reply to: