Re: glibc 2.3.1 status
At Mon, 20 Jan 2003 20:44:40 -0600,
Anthony Towns wrote:
> On Sat, Jan 11, 2003 at 02:05:20PM +0900, GOTO Masanori wrote:
> > At Fri, 10 Jan 2003 18:23:32 -0500,
> > Daniel Jacobowitz wrote:
> > > On Fri, Jan 10, 2003 at 05:02:55PM -0600, Anthony Towns wrote:
> > > > Hi guys,
> > > > So, it looks like the arm problem's been worked out, thanks to Phil
> > > > Blundell. You've still got problems with at least:
> > > > * #175511: threaded errno broken on alpha, fix in CVS
> > Thanks Phil and Daniel :)
> > BTW, I would like to know how to solve arm problem. Is it buildd
> > issue?
>
> It was a kernel/glibc issue -- the new glibc used the ABI in a way that
> used to be different and the 2.2 arm kernels had some "compatability" code
> to make it work the old way, breaking the new glibc. AIUI. Fix is to remove
> the compatability stuff from the 2.2 kernel and/or to move to 2.4 kernels;
> which was also why the test build worked fine.
Ah, I see.
> > > > * #175526: glibc 2.3 breaks gcc 3.2 tests on m68k
> > > > * #175529: re_compile_pattern broken
> > #175526 is now under building and checking. I was surprised that
> > gcc-3.2 could not build with debian gcc-3.2 latest on m68k. Phil
> > suggested me some hints (use gcc-2.95, or -O), so I'm trying its test.
> > m68k is too slow to build, it's needed more time.
The current status of my gcc build on m68k is: still building.
It has been built for 10 days. At least it needs some more days.
The halfway of this result is that the current combination of both gcc
and glibc does not break so many tests. The problem is that the
current m68k gcc cannot build itself (I use gcc 2.95 for 1st stage).
> So...?
So, #175526 is not critical bug, and it may reassign to gcc-3.2.
> > I guess #175529 is related with another regex (mutt error) problem.
> > Is Someone tackling this issue? I plan to contact this report to
> > regex upstream author Hasegawa-san.
>
> Likewise?
Well, from BTS, the regex problem may fix. I'm testing it.
> > Well, me too. BTW, is this weekend appropriate? Due to gcc-3.2
> > transition is under going, next weekend is more suitable even if we
> > fix all bugs?
>
> There're enough real problems to worry about, without delaying for potential
> problems. If it turns out more stuff needs to be done, it can be done in an
> additional upload.
That's right.
OK, I plan to release -10 after checking m68k and regex, ASAP.
AFAIK, hppa (kernel version check), alpha (threads problem fixed by
Daniel, #175511), conflicts with wine (#170385), VIA C3 issue for a
long time, are already resolved. I wish 2.3.1-10 really untangle much
libc6 complaint.
Regards,
-- gotom
Reply to: