Re: [Jack-Devel] ALSA PCM substreams

On Tue, Dec 3, 2013 at 11:51 PM, Gabriel M. Beddingfield <[hidden]
> wrote:

> I don't think independent.  From inspecting the code... if you open up any
> substream and set the rate -- all other substreams (playback and capture)
> will then be constrained to that rate.
>
> So... maybe this would make sense...
>
>    hw:n,0,0 and hw:n,0,1 - "stereo pairs" (legacy)
>    hw:n,1 - "multichannel"
>
> ...and hw:n,1 and hw:n,0 would be mutually exclusive.
>
> It wouldn't "just work" with JACK... but a little googling would return
> "use hw:1,1" instead of "open ~/.asoundrc in an editor...."
>

I faced a similar dilemma when I first ported the existing RME hammerfall
driver to ALSA (way back in 1999). The windows driver had the option of
being accessed as a series of stereo pairs, and that was my first design
model. It turned out to have a lot of overhead and wasn't very useful. It
also didn't fit into the basic ALSA model of hw:* representing raw hardware
capabilities.

The problem here is that the level of independence of these streams is
fuzzy. Start/stop is independent, but sample rate is not. I would make a
guess that the period size may also not be independent. I don't think that
ALSA provides a clear rationale for one or the other here, but Gabriel's
suggestion seems smart to me.

1386162753.26607_0.ltw:2, <CAFa_cKkGTfnb0WO6N90rVhB1vjqqspJW9WsK5T7sfiKt+s=sgQ at mail dot gmail dot com>