On Monday 16 July 2007 22:37, Frans Pop wrote: Breakage in daily builds ------------------------ > As some of you may have noticed, daily builds have been failing all > over the place recently. Most issues are now resolved. The zlib1g problem should as of today also be fixed for arm. There is still one issue left that (builds are OK, but installs fail because of a missing symbol in reduced libc) cannot be resolved until a conflict between libc-dev and linux-libc-dev (IIUC) is resolved (#434040), but fortunately that issue only affects a few architectures. The issue also means that some udebs may FTBFS when uploaded. Kernel status ------------- > Architectures that are still at older kernel versions: > - powerpc (2.6.18): could be switched; Colin has now done the upload so powerpc has been switched to 2.6.21 too. > - hppa (2.6.18): there are fairly big kernel/toolchain problems for > hppa that have made it fairly pointless to switch; as 2.6.21 is not in > testing anyway, it probably makes sense to make the switch for D-I as > well and just accept any brokenness Uploaded yesterday, waiting in new. > - sparc (2.6.20): sparc32 was disabled in 2.6.21 and I've not wanted to > disable it in D-I/debian-cd As dropping sparc32 has now been officially announced, I've updated the kernel for only sparc64 and disabled sparc32 in the daily builds. Other issues ------------ We've had quite a few reports about installations hanging during task installation. Colin has today traced this to a nasty race in debconf and uploaded a new version that will hopefully fix this. After we've had some confirmation that the issue is now really resolved, this fix should be backported to Etch for the next point release (probably 4.0r2). > In principle we could do a D-I release based on 2.6.21, but current D-I > has not really been tested (because of problems with daily builds) and > there are a few issues that should be fixed before a release: > - CJK issue in slang2: #425835 > - regressions in cleaning old LVM partitions: #425829 Both unchanged. > Steve Langasek recently traced the issue that's been affecting alpha > (builds fine, but result is unusable). A minor patch is needed in > mklibs (#433388) so that needed symbols with type STT_NOTYPE are > recognized during library reduction. Fixed. > D-I builds for mips and mipsel are completely broken and basically have > been since shortly after Etch was released. I understand there is slow > progress, but have no idea what current status is. Unchanged.
Attachment:
pgpwgSrHcuKq6.pgp
Description: PGP signature