Re: Building the whole Debian archive with GCC 4.1: a summary
Le Samedi 25 Mars 2006 22:54, Martin Michlmayr a écrit :
> Obviously, compiling the archive from source leads to many important
> insights. What I found out over the last two weeks is just how much
> work it actually is. We should try to create better infrastructure
> to make manual compilation of the whole archive easier. It seems
> that Roland Stigge has done some work on this already with his DARTS
> project [5] but I have yet to take a look at it.
I've been working on a Perl script to use pbuilder to do a recompilation of
the archive off and on for a few years now. There are a few packages, such
as gnat, which obviously cannot be compiled from scratch, but otherwise I've
managed to create an accompanying set of "cycle-breaker" scripts to work
around circular Build-Depends, so that bootstrapping from just the
build-essential packages should be possible. On the other hand, these break
frequently, and it would be nice if the involved packages could somehow
support doing the reduced initial builds; but I haven't come up with a
reasonable way of doing this yet.
I've considered packaging the script as "pbuildd" several times, but at the
moment setting up the build system is a lot of work, and the script itself is
not very configurable (e.g. you'd have to modify it by hand if you wanted to
try building testing instead of unstable, or include experimental and/or
non-free packages). I guess an alternative is to put my work up on an alioth
project, or maybe somewhere under the pbuilder alioth project, if there's
interest.
By the way: my latest experiment is putting in a hook script to build
everything as if it were a binNMU, with a version of e.g. 1.2-3+pb1, in order
to help apt distinguish those versions from the official ones. The good news
is that most packages build with no problem like this; on the other hand, a
few of them FTBFS or break. What would be the appropriate severity for such
bugs? I'm leaning towards "serious" myself, since the release team should
expect to be able to schedule binNMU's without worrying whether they'll work.
(And I'm impatiently waiting for XOrg 7.0 to enter unstable, to see how that
will affect the number of cycle-breaker scripts needed. :)
--
Daniel Schepler
Reply to: