On 2015-04-08 09:34, Takashi Iwai wrote:
At Wed, 08 Apr 2015 09:20:56 +0200, David Henningsson wrote:
On 2015-04-07 18:06, Takashi Iwai wrote:
This would work, yes. But, I have some uneasy feeling, something not well digested...
Ideally, we want a single API for representing both input and kctl jacks.
Maybe this is somewhat my fault for steering Yang in that direction. But the requirements are somewhat different.
HDA has the phantom jacks, and the exact naming for each kctl requirements. ASoC has the combination/button requirements, i e one jack can represent more than one kctl.
The phantom jack requirement means that the snd_kctl_jack_new API cannot be removed straight away; we could move it to be internal to HDA (it's not much code anyway), but I don't see a need for that.
But the HDA code can be moved around to look like this:
if (phantom_jack) { snd_kctl_jack_new(); } else { snd_jack_new(); snd_jack_add_new_kctls(); }Now the HDA looks more like the ASoC variant. Yang, what do you think about that? That would make the API simpler, wouldn't it?
Well, what I thought was rather to allow snd_jack_new() creating a phantom jack, too, with some flag. When a phantom flag is set, it creates no input jack device but only kctl jacks.
Yes, that sounds even better - that would make it easy for non-HDA drivers to create phantom jacks too, if they want.