glibc 2.3.2-2 goes unstable
Hi all,
Our long journey for both uploading glibc 2.3.2 to unstable and
cvs.dpatch update to the newer one from Dec 2003 has just arrived a
hopeful important corner. I dupload 2.3.2-2 for unstable after a bit
hours. I believe this 2.3.2 becomes the base library for releasing
sarge.
This upload is triggered by suddenly uploaded 2.3.2-1 into sparc
unstable. We should solve this confusion quickly.
AFAIK, at least almost all architectures (alpha, arm, i386, ia64,
m68k, mips, mipsel, powerpc, s390, sparc) are build ok with upstream
2003-07-08 cvs (not current debian-glibc 2003-07-15 cvs, but I think
no big change is introduced). Only hppa is not ready, even building,
so this is not complete update for sarge. We all cheer Carlos
O'Donell and hppa guys with their extraordinary work and effort.
I heard from jbailey that unreleased architecture hurd-i386 was not
OK. It's hard task to fix because they need to modify for newer
kernel L4. I don't know sh status, but I expect it's not hard to
support because of Kaz's great jobs.
I added below changes for Jeff's 2.3.2-2:
- debian/libc/DEBIAN/postinst: Modified watermark of restarting NSS
services from 2.1.94-1 to 2.3.2-2, because libnss_compat is changed in
2.3.2-1 (2003-06-17 Upstream change).
- debian/patches/80_glibc232-wcsmbs-fix.dpatch: Added to fix wcsmbs bugs
which is lacked in 2003-07-15 upstream cvs. (Closes: #202969)
- closing #201221, and #186331.
We can fix RC bugs in this release:
* #202969: libgtk2.0-0: Crash with gtk_file_selection_get_selections
I put patches in the debian-glibc cvs.
* #186331: alpha static adjtimex
I tested and found this bugs were fixed in Jun 2003.
* #201221: Bug in dlopen/dlclose leads to segfaults with kdecore
We know this problem is fixed in 2.3.2 based glibc.
The remained RC bugs are:
Critical bugs - outstanding
* #203322: python2.2: Python fails with illegal instruction during postinst on sparc32
* #203324: libc6: __strtod_internal fails with illegal instruction on sparc32.
I would like to know BenC's thought. Actually this is not
only glibc problem.
* #203303: glibc: gcc3.3 complains swab.h fails to conform to ISO standard
* #203949: libc6-dev: gcc3.3 complains that ptrace.h on s390 fails ISO C
They are kernel header problem. If they are still remained in
2.3.2-2, then I reassigned them into other package and ask to
fix.
Grave functionality bugs - outstanding
* #184048: [m68k] binutils testsuite failures built in a glibc-2.3.1 environment
I checked but some tests were still unexpected failures. I
would like to hear the opinion especially from Matthias Klose
and Daniel Jacobowitz.
Serious policy violations - outstanding
* #181493: glibc: Sun RPC code is non-free
* #181494: GNU Free Documentation License is non-free
I discussed with jbailey in debconf3 that these bugs should be
downgraded. Because these bugs are difficult to fix quickly,
and actually these are not only glibc problem but also other
programs which include RPC code or GFDL documents. BTW, do we
have a consensus for packages which hold GFDL documents?
So, from the analysis of these RC bugs, the remained important bugs of
libc6 2.3.2-2 which need to fix quickly are #184048 (m68k binutils)
and hppa build failure.
I put all compiled .debs (i386) at
http://people.debian.org/~gotom/glibc/2.3.2-2/
Or apt-line
deb http://people.debian.org/~gotom/ glibc unstable
Regards,
-- gotom
Reply to: