>>>>> On Mon, 14 Dec 1998 14:12:34 +0000, "P.J.Leonard"
<[EMAIL PROTECTED]> said:
> Thanks that seems to do the trick. I will now attempt to decouple the
> patch
> management from my sequencer program so they may be more reusable.
Another merit of this sequencer is that the user can send
real-time control events to the queue (not by using ioctl).
In the next release, the MIDI input will be also shareable, and input
and write queues can be redirected between clients.
> This does lead to some questions about interprocess communication. One
> of the
> nice things about my program at present is the user need not be
> concerned with
> bank/preset numbers (the patch management just uses the current voice).
> I also
> save the loaded patch info and automatically restore it when a song is
> read in.
> If anyone has any views on how sound applications should
> communicate/who owns
> what etc then I would be interested in your opinions. My present plan is
> to use
> QT and use sockets to communicate ?
It's possible to send packets through sequencer device. However, if
the communication should be processed in real time, the socket would
be better than queueing through sequencer. The sequencer is suitable
only to process the events on the schedule. (So, the patch loading is
not suitable for "sequencer" device, originally...)
Maybe, this sequencer code can be ported also on ALSA.
Indeed, I'm already using some functions defined in ALSA, because of
good design and also of compatibility.
Then, some existing drivers (including awedrv) can be ported without
much difficulty.
--
Takashi Iwai / [EMAIL PROTECTED]
Department of Materials Science
Friedrich-Alexander-University Erlangen-Nuernberg
