New POSIX semaphore functionality

Corinna Vinschen
Mon Oct 27 12:58:00 GMT 2003

On Mon, Oct 27, 2003 at 11:35:27PM +1100, Robert Collins wrote:
> On Mon, 2003-10-27 at 23:32, Corinna Vinschen wrote:
> > IMHO it would be a pity to require cygserver for POSIX semaphores.
> > Except for sem_unlink(), every other functionality neatly matches
> > what is allowed by Windows.  Perhaps we should implement sem_unlink()
> > so that it just returns -1 and sets errno to ENOSYS for now.
> Well, I don't see the cygserver as a bad thing - setup as a service,
> it's no more intrusive that RPCSS, for instance.

It's not a bad thing, but it's a bad thing to require an additional
running service for stuff which could be implemented without it.
POSIX semaphores are such a thing.

> And for the semaphores, what I was thinking was that the cygserver
> allocates anonymous semaphores, handing out the handles - not that the
> cygserver would be used for each and every call.

Which is not required.  Anonymous POSIX semaphores are working fine,
don't they?  Only named semaphores and only together with sem_unlink()
are a problem.

XSI semaphores are an entirely different problem, of course.


Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Developer                      
Red Hat, Inc.

More information about the Cygwin-developers mailing list