[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: apt-get upgrade dies badly ...

On Thu, 28 Oct 1999, Paul Slootman wrote:

> I've not yet got to a stage where my local mirror is complete enough to
> attempt an upgrade; last time was a couple of months ago (from
> experience I've discovered that upgrading unstable/alpha over the net is
> not recommended...). However, does this problem only affect a couple of
> people, or anyone who installs that libc? If only a couple of people,
> one would say there must be something in their setup that triggers it...
> Hmmm.

I believe it's really related to apt since the two so far that have had
problems tried with apt.  Maybe a static bash and apt would be in
order...I know, we've had arguments on -devel about static binaries, but
this makes sense until I can get libc to cooperate with apt again.

> As you may have noticed, mutt when compiled with -O2 triggers a compiler
> bug (a function is called with an array value, and the array value x-1
> is actually passed).  Do you use -O2 for libc? (although I could
> reproduce the problem with gcc 1:2.95.2-0pre1.1).

Yes, but I can't use the normal packaged gcc to compile libc since it
still has broken complex support.  I used the cvs version of egcs (with
the patch to fix the imaginary accessing with complex numbers).  I'm going
to try extract the imaginary patch from the repository, though, to see if
we can just squeeze that patch into our normal gcc for now.

> Unfortunately gcc is still not quite there for alpha...

I agree.

> BTW, why is bigloo-dev not uploaded? bigloo is, and it's built from the
> same source apparently. I also couldn't find bigloo in the autobuilder
> status (Chris, when you update the webpages again, please add a link to
> http://www.complete.org/~buildd/ ).

Oh, good idea (forgot when I did it yesterday...I've had alot on my mind
the past two days).  Unknown as to why bigloo-dev isn't up.


Reply to: