(Reopen) "always_on" not working for ALSA devices

[Cleric-K]

I'm not sure if #1599 gets attention since it is closed but the always_on option currently doesn't work for ALSA. I have described in the last post of the issue that simple change in the code makes it work. Is it intended that ALSA doesn't support always_on? From my experience everything works perfectly well with the above mentioned code change.

[GK-774]

Yes, correct "always_on"does not work, reports as "closed". If this cannot cannot be controlled from within mpd it would preferable to remove the option.

[Cleric-K]

Yes, in the way the code is compiled right now it does not work. But as explained in the linked issue, by modifying AlsaOutput::Pause() to return true, everything works great, as far as my tests have gone. So my question is why don't we simply change the return value of the method and make always_on work with ALSA? Is there something I'm overlooking? Maybe something that will break down in this way, something that I haven't covered in my tests?

[GK-774]

If you simply return "true", always_on "no" (or having it not defined at all) would not work, it would be "on" in any case, no? Or have I misunderstood this?

[GK-774]

Please disregard my last comment. Just tried it and yes, this makes it work as expected.

[GK-774]

Well, no. It keeps the output open but consumes 100% cpu when stopped. With always_on "no" it behaves normal. It´s probably not a good idea at all at least not for an USB audio device.

[Cleric-K]

Hm, you are right. Since I use the device only for audio player I haven't noticed the high cpu. I'll try to look through the code if I can spot something. Most certainly the high cpu comes from some code that keeps looping as fast as the cpu allows.