Re: X Strike Force X.Org X11 SVN commit: r385 - trunk/debian
On Thu, Jul 21, 2005 at 07:06:59PM +0200, David Martínez Moreno wrote:
> El Jueves, 21 de Julio de 2005 18:00, David Nusinow escribió:
> > Nope, sadly -4 is in the archive already since I messed up a couple of
> > things in -3. -4 is in good shape so far though, so it may be able to go in
> > to testing. While I'm tempted to get -5 with your SELinux fix in first, I
> > think unblocking things like gnome from migrating to testing is probably
> > more important.
>
> David, that is impossible, as xorg-x11 has several grave and serious bugs as
> today. We should fix them before. And the other option, to downgrade the bugs
> or manually force the release to testing, ends in the same blocker: the
> release team. They have the last word.
I'm less concerned about the debconf bug, and may well downgrade it if I
don't fix it first, but the alpha ELF bug does need to be fixed. The ELF
bug also highlights that we *need* to get working on 6.9/7.0 packages as soon as possible.
> Do not forget the gcc-4.0/libvgahw.a bug as well. I would like to ship
> well-built code in testing. As Eugene stated a couple of days ago, there are
> spreaded volatile's all along the code, not only in libvgbahw.a.
We have a workaround that I'm willing to put in to testing this early in
the release cycle in order to unblock vast portions of the archive. Also,
the gcc maintainers will upload a fixed version with a backport for our
problem as soon as their current version migrates to testing.
> I think that we should concentrate on current unstable release, now that
> (hopefully) compiles on every architecture, and we are not blockers for KDE
> or GNOME or whatever other graphical package.
We are currently blocking any graphical package from migrating to testing.
If we don't get X.Org in to testing, it's going to be a mass migration of
these things in at once, and I'd rather try to avoid that if possible.
> Migration from xfree86 packages to xorg ones are far from perfect. I strongly
> think that we must keep the number of users small until we achieve maturer
> packages.
I'm not too concerned about this for testing. It definitely needs to be
fixed, but it's not a blocker to migrate to testing.
> Our priority now would not have to be to cause that the packages arrive at
> testing, but to leave them in unstable some time until the things calm.
I'm not going to worry about polishing these packages until they're
perfect. I actually have no intention of shipping them with etch, in favor
of either 6.9 or 7.0, depending on how the transition goes. I want very
much to unblock testing so that half of Debian isn't waiting on us.
All together though, due to the ELF loader bug, it'll likely be -5 which
I'll let go in to testing, as it'll fix the gcc4 issue (we'll wait on the
gcc team to build and upload) as well as the alpha issue if Steve's patch
tests out Ok. I'd like to sneak the quilt-based patch system that Eugene
has worked up in that upload as well, and possibly a few more items on the
TODO list (which needs to be more carefully prioritized).
- David Nusinow
Reply to: