[arch-dev-public] request for toolchain signoffs

[Andreas Radke]

Am Sun, 30 Nov 2008 15:08:54 -0600
schrieb "Dan McGee" <dpmcgee at gmail.com>:

> > No major problems with glibc, but should we move this, while we know
> > this breaks DNS resolving for people that have a broken DNS server?
> > I solved it at the office by installing pdns-recursor and using
> > localhost as DNS server, but if this glibc reaches the next
> > installer ISO, it's not so easy to install pdns-recursor there,
> > which breaks network installations.
> 
> I was of the same mindset here- does it kill us to sit on this and
> wait for a fix? I'd prefer to not close 5 bugs a day saying "won't
> fix", and signing off and moving broken stuff to core seems pretty
> sloppy of us from the end user's perspective.
> 
> -Dan
> 

I've sent a reminder to the glibc list but I don't expect any (helpful)
answer. We have the same state of checkout that is used in Fedora 10
release and in their devel branch. You know glibc is maintained mainly
by RedHat/Fedora.

http://sources.redhat.com/cgi-bin/cvsweb.cgi/libc/NEWS?rev=1.196&content-type=text/x-cvsweb-markup&cvsroot=glibc

Version 2.9

* Unified lookup for getaddrinfo: IPv4 and IPv6 addresses are now looked
  up at the same time.  Implemented by Ulrich Drepper.

To me it seems there isn't much we can do here. And we will hardly
see any fix other than to rollback the glibc. The more glibc2.9 will be
used the harder will the pressure become to dns server admins to fix it.
Afaik OpenSuSE will also use glibc2.9 in their upcoming release.

For me it's not a showstopper. Maybe it's worth a new item though.
Opinions?

-Andy