Bill Studenmund <wrstuden@nas.nasa.gov> writes:

> I'd tend to vote for #3 (POSIX), but I'm not in a position to do anything
> about it. :-) I think you should go with #2 unless someone offers to do
> #3. Just mark it as "most likely going away".

I'm likely going to go for #2 and mark it as "going away."  IMHO,
having an interface that is so marked and working pthreads is MUCH
better than not having them and doing serious black magic in the
pthreads implementation with timers.

This won't get around the issue with SIGCHLD, which our version of
pthreads also uses.  But, that one signal is far, far easier to deal
with than all the others.

Before I get set on adding a syscall, however, I'm going to see about
other methods.  I should be able to find one...

Question:  How do I arrange for a function (say, pthread_init()), to
be called before main()?  And how does that work when it may be a weak
reference, so pthread_init() can be replaced at link time?

--Michael