On Sun, Oct 11, 2009 at 09:59:33PM +0200, Samuel Thibault wrote: > Remi Denis-Courmont, le Sun 11 Oct 2009 12:11:50 +0200, a écrit : > > ----- Message d'origine ----- > > > Here is a patch that adds only pthread_condattr_init/destroy, > > > pthread_cond_timedwait, pthread_exit, and makes both cond_*wait abort > > > instead of just returning 0. > > > > I would expect cond_timedwait to sleep for the specified interval rather > > than abort (though sleeping in a portable way is not completely trivial). > > No? > > The problem is that we don't have an example where an application > would actually need that behavior. If the application is waiting for > something that can only be signaled by another thread (that's what > condition variables do) while threads are not actually used, there's > probably something wrong in it and it's probably better to just report > it accurately.
I can easily imagine an application that wants that behavior: "wake up after X amount of time, or sooner if awakened by some other thread". However, I do believe we should wait for an actual use case before supporting that. - Josh Triplett -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
