This is the mail archive of the
mailing list for the glibc project.
Re: Undefined reference to atexit (FAQ).
On Thu, 8 Nov 2001, Martin v. Loewis uttered the following:
>> Is there any prediction on how long it's going to take to fix this ? I
>> think gcc and glibc development should be more integrated. It's not the
>> first time something like this occurs.
> I don't think they should be more integrated.
Indeed much of the delay and argument here is in trying to get glibc to
fix this without including big chunks of GCC or otherwise becoming
dependent on particular GCC versions.
> Each package is quite
> complex on its own already. If you would, e.g., tie the release
> schedules of the packages together, release would take much longer.
> So you would have a perfect release, but you would have to wait five
> years for them.
And this would violate the GCC release schedule ;} ;}
> Furthermore, gcc is not just for Linux and glibc. None of these
> problems exist, say, on Solaris. Why should Solaris users suffer from
> problems they couldn't care less about?
This is why the GCC team would sooner burn than integrate GCC into
glibc. glibc is dependent upon GCC, but not vice versa, and almost
everyone on the GCC team has a pleasingly platform-neutral
attitude. (How else would people from a dozen Unix vendors and rival
free software groups cooperate but being reasonable about platform-
`You're the only person I know who can't tell the difference
between a pair of trousers and a desk.' --- Kieran, to me