[fedora-arm] rawhide build status
I said last week that I would drop an email to the list with details
of the plan for rawhide building.
Current blocking items are follows:
1) Problems building core build root/chain packages in koji (gcc,
python, coreutils, findutils, gettext, libssh2)
2) gcc upstream patches needed for build
3) other unknown core packages with non upstreamed components
1) The problems building the core packages mentioned about were
primarily either unknown or pending an upgrade on the koji builders.
Having done some testing (thanks Dennis) we're now fairly confident
that an upgrade of the builders to a consistent image/kernel based on
F-15 should fix most (and likely all) our current problems. Not to say
there's not other issues we'll run into but I'm quietly confident
we'll be able to move forward
2) ggc: blocking on PR 51442 [1], would be good if we could get
forward movement upstream on this.
3). Other packages. I'm aware of a few other problematic packages that
will need fixes but none are really what I would call core that should
stop a mass rebuild from kicking off. I plan to deal with these
break/fix style. There have already been a number of fixes I'm
committed to rawhide during the F-15 initial upstreaming and
subsequent building.
[1] http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51442
Moving forward for the plan to build rawhide. The process will be
simply build the core buildroot packages and then run koji-shadow.
Along side koji-shadow I'll likely do some ordered/targeted dep chain
building. The problem with rawhide over the current F-15 building is
that we really need to build things in order as we don't have a
pre-built F-15 that we're just rebuilding in koji so we need to be
aware of and take into account ABI and soname changes to ensure we
don't end up with a broken infra.
The basics is:
1) Import f16 noarch packages (alot of f17 is still f16 and even f15) - done
2) Import a snapshot of f17 noarch packages - done
3) build an ordered build root - reviewed and in progress (problems 1
and 2 above blocking)
4) run koji-shadow
5) break / fix
6) get upstream teams that are interested to build their stacks as we
move further into the build (haskell, kde, xfce etc)
7) target initially a core minimal rootfs and move out from there.
8) onwards an upwards
9) f17 branch etc....
All as fast as possible!
Currently we're awaiting the Seneca guys on the much anticipated infra
upgrade and the gcc upstream bug. My thoughts are to do a local arm
gcc build with the patch so as not to block short term progress.
I look forward to thoughts and feedback.
Peter