Alright -- I think I've got it working.  After mucking about for a bit
with asyncs, I realized that it makes more sense to evaluate cleanup
handlers in do_thread_exit -- and evaluation should happen there
anyway, since pthreads says cleanup handlers get called when the
thread exits for any reason, not just cancellation.  Duh.

The patch I've attached adds three new public functions:
cancel-thread, push-thread-cleanup, and pop-thread-cleanup.  API
documentation on their behavior is included in the changes.

I've never submitted a patch for Guile before, so I've likely made a
few mistakes in formatting, etc., and I don't really know the vetting
procedure. I hope I've gotten the main bits right, though.  Please let
me know if there are any changes I should make.


On 9/20/07, Julian Graham <joolean@gmail.com> wrote:
> > Would it be possible to defer execution of the Scheme code (cleanup
> > handlers) to after the C cleanup handler has been called?
> >
> > I.e., the C handler would push the Scheme handlers to a list that would
> > be eventually executed, when Guile is back into a "clean", regular
> > state.
>
> Yeah, that would be fine -- except that after the thread is signaled
> for cancellation, it'll never go back into a regular state.  That is,
> I think the next thing that happens is that the thread-local exit
> handlers (on_thread_exit in threads.c) get called -- and I don't think
> that code evaluates queued asyncs (please let me know if it actually
> does!).  And any queued asyncs would have to be evaluated within that
> thread (and in order), because people are going to want to do things
> that have to be done from within a particular thread, like unlock
> mutexes.
>
> > I'd prefer thunks as well, it looks more Schemey.
>
> I agree -- and makes things way easier if it turns out it's possible
> to do this with asyncs.
>