Re: glibc bugs
Hi,
I'm one of maintainer debian-glibc.
At Thu, 27 Feb 2003 20:26:44 -0600,
Steve Langasek wrote:
>
> [1 <text/plain; us-ascii (quoted-printable)>]
> On Thu, Feb 27, 2003 at 09:18:37PM -0500, Matt Zimmerman wrote:
> > On Thu, Feb 27, 2003 at 10:52:52PM +0100, Josip Rodin wrote:
> >
> > > On Thu, Feb 27, 2003 at 09:59:58PM +0100, Otto Wyss wrote:
> > > > Since I woundered why glibc doesn't move to testing I've looked at the
> > > > bugs if I could help. But than I woundered even more that these
> > > > licensing differences can't be solved within 85 days. Can these 2
> > > > serious bugs somehow fixed, downgraded or closed?
> > >
> > > They are de facto downgraded by their set of tags. The real problem with
> > > glibc are other things, like those listed at the update_excuses page
> > > (excerpt available from glibc PTS web page).
>
> > Said excerpt lists two excuses:
>
> > 1. "taking my time about glibc", which I assume is a manual override, and
>
> > 2. glibc (source) is buggy! (2 > 0)
>
> > So it would seem that these bugs are not being disregarded (even if they are
> > not the only barrier).
>
> However, AIUI, this is mostly a cosmetic bug in the testing scripts (or
> a metabug in the way the bug is tagged). It's my impression that, if we
> get to the point where glibc really builds and works on all archs, the
> intention is to ignore these bugs as pre-existing and move glibc into
> testing. This is somewhat based on hearsay, but it also jives with my
> understanding of the testing process in general.
Well, such two excuses are old. In addition, RC bug (#181493,
#181494) is license issue, not program bug. Jeff Bailey and me with
upstream plan to fix it in future.
The problem remains in debian-glibc is only libgcc-compat symbol issue
(#179781 and something). It's much stable (yes I know there are some
bugs as "important", but anytime a program has a lot of bugs :-).
libgcc-compat things is as follows: In the transition from 2.2 to 2.3,
some symbols become as hidden and not resolved related with libgcc
symbols. Soo it's possible that upgrading from to sarge makes some
packages unusable. Guido Guenther and me (alpha, arm, ia64, m68k,
s390), Carlos O'Donell and Randolph Chung (hppa), plus some people
(ex. Ryan Murray, Daniel Jacobowitz, ...) are working with this issue.
Anthony Towns judged that it should be fixed. IMHO, it's not needed
to fix this bug for putting glibc for sarge. This problem affects
only some limited packages and user-self-built software. But yes it
should be fixed in early days. (Though, I'm afraid you may complain
"oh, some binaries becomes unusable, debian-glibc team s*cks!" :-)
So I think it's OK to release glibc for sarge in next 2.3.1-15, but if
you want to make sarge as "stable", please wait for some moments.
Regards,
-- gotom
Reply to: