[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [decision] new patch management system in xorg-x11



On Sun, Jun 05, 2005 at 10:54:57PM -0600, Bdale Garbee wrote:
> david_nusinow@verizon.net (David Nusinow) writes:
> > The modularization effort seems to be progressing,
> > but if we can move to quilt rapidly then we'll be all the better for it,
> > since that experience can be applied to the modular tree packaging when
> > it's ready.
> 
> Why not move directly to packaging the modular tree?  Time spent working on
> putting a monolithic x.org build in Debian seems like wasted time to me.

The modular transition is very tricky -- indeed, large swathes are not
completed.  The modularisation of the server right now sits only in a
directory under ~daniels on my laptop.  Most drivers aren't done, and
not of all the libraries are even completed.

Throw in a /usr/X11R6 -> /usr transition (mindblowingly difficult in
myriad ways, as I'm discovering), and you have something which we cannot
move to right now.

I advocated moving to it earlier, which was fine; it's still a very
worthwhile project to sit down and plan for.  Josh Triplett and I are
doing that now, and we're rapidly approaching a fully-converged package
set for Debian and Ubuntu, which is great.

But, the reality is, that's still not right around the corner, and we're
stuck in this ridiculous situation where people who bought computers in
mid-to-late 2003 can't run Debian released in mid-2005, which is
absolutely, mindblowingly, absurd.

We need X.Org in unstable, and we need it in unstable ... well, early
last year.

We have the package set from Ubuntu already (personally I'm phenomenally
skeptical of the package audit, and see no purpose except to prove a
point, but if it has been deemed by the lead X.Org maintainer that this
is the process that must be followed to get it into Debian, so be it --
after all, he is the one doing all the work), so importing that into
Debian and using it there should be relatively trivial.  It already
builds and runs fine across amd64, hppa, i386, ia64, powerpc, and
sparc, and I'm working on porting it to alpha, arm and mips personally.
Any porters who are interested in porting it to their platform should,
again, contact me or the list.

> Helping drive completion of the work to make packaging the modular tree 
> possible is the sort of stuff we're collectively pretty good at, and it'd be
> great to put Debian in a "leadership" position in the larger X community.  
> 
> Thoughts?

Debian has no credibility in this sense; we're so abysmally far behind
now that all our previous credibility has been shot to pieces and we're
a laughing stock within the X world.  I think we need to make the
incremental step to a halfway-recent version of the monolithic tree
first, while myself and Josh continue our work on the modular tree, so
we're out there and ... well, at the rate Debian moves, not too far
behind.

If the current procedures must be followed through, I estimate we can
start making a concerted effort to get the modular tree into sid in
mid-2006.  I will, of course, continue my non-Debian development on
this, and I'm sure it will be available in many places and likely
backported just as xorg is widely backported now, just as xfree86 4.3
was widely backported from 4.3.0-0ds*, just as xfree86 4.2 was widely
backported from Ishikawa's packages.

I think the best we can hope for is that more people join myself and
Josh in the modular packaging effort, so even if we don't get the
modular tree into sid for various reasons for a very long time, we can
at least have very visible external development, and a viable non-sid
source of X development.  Obviously, this will require much more
effort and discipline to make it viable than development in sid (or even
experimental), but given the stated position and previous actions of the
current controlling interests of X in Debian, I don't see any other real
way.

In short: if we route around the current horrendous breakage that is the
XSF, we have half a chance.  If we continue to dance around with
ludicrous patch audits while users who cannot start X twiddle their
thumbs (or, far more likely, go somewhere else), then the situation is,
as it has been for the last four years, utterly hopeless.

Cheers,
Daniel

Attachment: signature.asc
Description: Digital signature


Reply to: