Test packages for libc6 2.1.91+cvs
I've gotten a set of packages using glibc 2.2pre from CVS. There are
binaries for sparc and i386, aswell as the source package. I'm interested
in several things:
1) Compilation success/failure on other archs
2) Testing of these packages, especially WRT, compiling other packages. DO
NOT file bugs on these packages. Send email to me. DO NOT file bugs to
glibc-bugs. They don't want anyone using this in production systems
anyway. I'm taking responsibility for this, and I don't want people
flooding them with trivial problems that I can filter properly to them.
A few caveats:
1) DO NOT, FOR ANY REASON, UPLOAD PACKAGES TO THE ARCHIVE COMPILED USING
I cannot stress this enough. If you upload a package that depends on
libc6 (>= 2.1.91-1), I will personally lart you :)
2) NSS1 compat libraries are not installed with this set. From what I have
been told, these allow you to use old libc5 conf files with libc6. If
you are one of the crazy ones that actually do this, don't install this
set. The NSS1 compat libraries from the 2.1.3 package wont compile with
this source, and I'm not sure it's worth getting it to compile.
3) The lpr package breaks with this. It is the only broken package I have
encountered so far. This is NOT a bug in glibc, it is a bug in lpd for
using __ivaliduser(), which is an INTERNAL glibc function, and not even
documented. Thus, lpr is broken. According to the comments in the glibc
source, lpr should be using ruserok() or iruserok(), and I have filed a
bug to this affect, but judging by the bug reports already filed on
lpr, I don't think it will get fixed. I WILL NOT hack around it in
glibc. This is what happens when you use undocumented things, and
especially when you have as crappy of code as there is in lpr. I looked
at it in the hopes of fixing it...and well, I wont do that again.
/ Ben Collins -- ...on that fantastic voyage... -- Debian GNU/Linux \
` firstname.lastname@example.org -- email@example.com -- firstname.lastname@example.org '