On Sat, May 06, 2006 at 12:29:53AM +0200, Esben Stien wrote: > Thomas Schwinge <[EMAIL PROTECTED]> writes: > > > translator to be installed on /dev/snd > > This was a bad design decision in the linux kernel, or so I > understand. Treating audio IO through a file descriptor is very > bad. The JACK[1] project use a callback model and is the only working > solution for realtime audio.
I don't really see the problem with file descriptors, could you explain it or give a relevant link about it ? Moreover, don't forget that on the Hurd, filesystem nodes are the links to Hurd services, which means we need at least one node somewhere to access the sound system (using Mach interface directly would be an awfully wrong design). Do you think it'd be possible to have JACK without an ALSA or OSS interface on the Hurd ? -- Richard Braun
signature.asc
Description: Digital signature
_______________________________________________ Bug-hurd mailing list Bug-hurd@gnu.org http://lists.gnu.org/mailman/listinfo/bug-hurd
