Re: opening control devices with "hw:N,M"

[Takashi Iwai <tiwai@suse.de>]

At Mon, 22 Jul 2002 21:21:47 +0200 (CEST),
Jaroslav wrote:
> 
> On Mon, 22 Jul 2002, Takashi Iwai wrote:
> 
> > At Sat, 20 Jul 2002 09:08:29 +0200 (CEST),
> > Jaroslav wrote:
> > > 
> > > On Fri, 19 Jul 2002, Takashi Iwai wrote:
> > > 
> > > > At Thu, 18 Jul 2002 09:03:26 -0400,
> > > > Paul Davis wrote:
> > > > > 
> > > > > why doesn't this work:
> > > > > 
> > > > >     snd_ctl_open (&handle, "hw:0,0", ...)
> > > > > 
> > > > > it means that you can't use the same standard name format for control
> > > > > devices as PCM devices, which is a bit of problem for programs that
> > > > > need to open both but only want to require the user to specify a
> > > > > single name ...
> > > > 
> > > > do you suppose that the second argument of ctl name is always ignored?
> > > > the control interface has no device like pcm.
> > > > 
> > > > but it's true that having a same naming rule makes the life easier...
> > > 
> > > The 'default' is there for this purpose. Otherwise, application / user 
> > > must know the right device name.
> > 
> > iirc, it's not easy to pass the card/device number to "default" pcm.
> > at least, "default:CARD=1" doesn't work.  the easiest way to access a
> > certain card is, so far, only to use "hw:x".
> > this must be fixed.
> 
> I don't agree here. The default is very simple case when the application 
> programmer doesn't want to take care about the right syntax of the device 
> specification.
 
ok.  so, you think that the default device is anyway exclusive,
which is not used for general type to multiple devices.

> Anyway, programmers should know that CTL devices are one per card. If 
> there is some real need to correlate PCM handle with CTL handle, then we 
> might create a special function like:
> 
> int snd_pcm_get_ctl_name(snd_pcm_t *handle, char **ctl_name);

it sounds not bad.
in the case of a multi pcm, it's not undetermined, though :)

btw, this reminds me another thing.
we have some controls which related with a certain pcm substream.
and so far we have no api to pass this relationship.

well, they are correlated via an index value.  yes, but confusingly, a
similar control can play a different role on the other driver.

perhaps by similar strategy, we may implement an api function for
that.

> 
> > also, i thought that ctl.hw is defined purely in
> > /usr/share/alsa/alsa.conf, and tried to redefine this to accept more
> > arguments.  but it doesn't change anything.  is ctl.hw hardcoded?
> 
> Extra things in configuration file are usually parsed as errors.

err, no, i meant that ctl.hw seems not to be read from the
configuration file...  most likely i made a stupid mistake.
don't take this serious yet.


ciao,

Takashi


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf