Re: Please refrain from non-RC gcc-3.4 uploads until a new version has entered sarge

[Matthias Klose <doko@cs.tu-berlin.de>]

Andreas Barth writes:
> * Matthias Klose (doko@cs.tu-berlin.de) [041217 14:40]:
> > [...]
> 
> Is this description for the release update correct?
> | Another minor change of the toolchain is pending. Gcc-3.4 started to
> | use libunwind, as the internal gcc version was buggy. In the end, this
> | turned out to block package transition to testing currently. The plan
> | to solve this is to include the libunwind-code into gcc, and work with
> | that. Also, a new glibc will follow, once the new gcc upload is finished.
> | The bad news is that the current status of this blocks (unrelated)
> | packages from going into testing, which is not too good.

In http://lists.debian.org/debian-devel/2004/12/msg01744.html Matthieu
Delahaye gives a nice summary. Maybe merge the both texts to:

Another minor change of the toolchain for ia64 is pending.  Gcc-3.4
started to use libunwind, but the the internal gcc version was
buggy. Therefore the libunwind shared library had to be built from the
libunwind sources.  The plan to solve this is to include the
libunwind-code into gcc, and work with that. Because we are not
allowed to introduce new base packages for sarge, the shared libunwind
library had to be included in libgcc1 and libgcc's shlib's dependency
had to be tightened to reflect this change. The unwind patches were
backported to gcc-3.3. Also, a new glibc will follow, once the new gcc
upload is finished.  The bad news is that the current status of this
blocks (unrelated) packages from going into testing, which is not too
good.

	Matthias