Confusion over audio device enumeration

[Martin Sandsmark]

On Wed, Sep 23, 2009 at 03:21:00PM +0100, Colin Guthrie wrote:
> I'm just adding some more intelligence to phonon/kde with regards to 
> pulseaudio (for background, see the end of this mail).

\o/

I'm no PulseAudio fan, but I can certainly see its use cases (networked sound
for thin clients, for example), so I'm very glad someone is poking into this.

> So the platform plugin (second link) provides both it's own devices and 
> ask the backend. But the backend (first link) also appears to ask the 
> platform plugin for it's devices as well as giving it's own too.

I think this is because it is supposed to configure ALSA too. But I haven't
looked much into that part of the backend plugin, so I'm not entirely sure.
:-)

> I quite like the concept of phonon and can see it's value from a cross 
> platform perspective (with quicktime backend on OSX and dx on windows 
> etc.) but am a little disappointed that there are several linux backends 
> - personally I'd prefer to see efforts concentrated on one backend 
> (personally, from a technical design and appropriateness perspective, I 
> believe this should be gstreamer but I don't want to get into a 
> religious war so I wont hammer the point!)

I couldn't have put it better myself :-)
I myself prefer Phonon-Xine, but mainly because Xine is more mature (as is
Phonon-Xine), and I am not particularly fond of GLib. But seeing as Xine
seems to be low-activity, it might be nearing its end-of-life.

But I'm completely with you on one-backend-to-rule-them-all, I don't see the
value of several backends, it just brings more work to maintain them all.
What could be interesting, though, would be to have a cross-platform backend,
like Phonon-VLC (which is maturing now).

> [...] But if we use pulseaudio in the way I am working towards, then *all*
> applications will honour this configuration (pulse can determine most
> applications' role automatically from the category in their .desktop file)

This sounds pretty nice, though I would obviously prefer if it was done by
mucking with the asoundrc, instead of needing a whole sound server, but
that's not what we're discussing here, so meh. :-P

-- 
Martin Sandsmark 
:wq