> Having used UNIX for a few years here in university, I am fully aware of
> devices being locked while they are in use. My distributed networks class has
> taught me the importance of that (we tend to do a lot of programming
> assignments dealing with file locking and IPC messaging).
>
> However, there is a feature that I wish Linux had that Windows (shudder) offers
> - multiple programs accessing the sound card at once.
Windows does not have this feature. The Windows sound driver can be
held by a single process, just like the Linux sound driver.
The difference is for Windows you write to the MCI API (or whatever
it's called this week) and it transparently mixes other streams for
you, before delivering it to the sound driver.
On Linux we have something similar. It's called the ESD, and if the
sound application community stopped using /dev/dsp directly, and in
the interests of users started using ESD, we'd all win.
ESD even includes a wrapper script that fakes /dev/dsp for all your
old legacy applications. You can get the functionality you want now
with no major effort on your part.
ESD has other potential benefits, including removing bloat like the
softoss wavetable driver and putting it in user space where it more
rightly belongs. ESD can also transfer sound streams over the net.
Despite this gushing about ESD, it does have flaws, but Linux would
be better off fixing ESD and using it rather than changing the UNIX
open() semantics just to make badly behaved sound apps work.
Take note that the exact same problem exists for a number of rather
important subsystems, including libggi vs X, and gpm vs X, and sane
vs sg. Some userland peace treaties would solve a lot of problems.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/