Sound reasoning

[Hans Petter Selasky]

On 02/08/15 22:48, Евгений Б. Спамаковский wrote:
> On Sun, Feb 8, 2015 at 10:29 PM, Hans Petter Selasky <hps at selasky.org>
> wrote:
>
>> On 02/08/15 22:22, Евгений Б. Спамаковский wrote:
>>
>>> Hi.
>>>
>>> Is there a point to writing a pcm device to have multiple "channel"s in
>>> the
>>> same direction?
>>> Say, two SND_FORMAT(AFMT_S16_LE, 1, 0) for record and two for playback,
>>> vs.
>>> just one SND_FORMAT(AFMT_S16_LE, 2, 0) in each direction.
>>>
>>> pcm_register() has so prominently two arguments 'numrec' and 'numplay' and
>>> even if they are ignored, one can do pcm_addchan() any number of times.
>>> How is this expressed to the user-land programmer?
>>>
>>> "Read the code" is a fine answer, but it would be nice if someone can
>>> readily explain the idea behind the options.
>>>
>>
>> Hi,
>>
>>
>
>> Sounds similar to what my "virtual_oss" is doing. See:
>>
>> svn --username anonsvn --password anonsvn \
>>        checkout svn://svn.turbocat.net/i4b/trunk/ports
>
>
>>From the description it looks like virtual_oss is trying to do what VCHANs
> already succeed at doing.
> No offense, but this still does not explain what the difference is between
> a pcm unit collecting its physical streams in one channel vs. having a
> channel for each...
>
> Let's say I'm five years old?

Hi,

Maybe I'm not answering your question ...

"virtual_oss" is a bit more advanced than VCHANs. It can also record 
what is played back all bit-perfect and clock synced. A bit more in the 
direction of jackd. A multi-channel audio device can be split by a set 
of configuration parameters into any number of mono/stereo or 
multi-channel OSS devices.

And there is a graphical QT based frontend: "virtual_oss_ctl"

--HPS