Ubuntu no longer works correctly without pulseaudio

Ben Klein shacklein at gmail.com
Tue Apr 28 22:52:16 CDT 2009


2009/4/28 Susan Cragin <susancragin at earthlink.net>:
>>wineoss is not a viable solution to the pulseaudio problem, but this
>>is a real nasty issue where the solution is "as long as it works for
>>you, it's fine".
>>
>>What exactly is the issue with purging pulseaudio? Though this should
>>be taken up with ubuntu's bugs tracker. pulseaudio should be
>>considered optional, since ALSA/dmix works wherever pulseaudio with
>>ALSA backend works, even if it is opt-out.
>
> Ben, sorry...
> You probably haven't been following this thread closely. I determined a couple of days ago that pulseaudio was not actually my problem even though I thought it was. I just kept the name of the thread.

If you're still using pulseaudio, and it works with setting winecfg to
OSS instead of ALSA, then it's the same old conflict between
pulseaudio and Wine, just that more than likely padsp or equivalent is
being used, and you're lucky enough to not have problems using it like
that.

> One of wine's developers asked me if linux's (or Ubuntu's) alsa-wrappers could have changed, and if wine's alsa-wrappers could therefore need updating.
> The fact that something works under winecfg OSS (but not under winecfg ALSA) is a valuable piece of diagnostic information.

Yes, but aren't we talking about pulseaudio here? Wine doesn't support
pulseaudio, so pulseaudio has to support Wine for things to work. With
most applications, pulseaudio's ALSA plugin (plug pulse) is suitable.
In the case of Wine and a few other apps, plug pulse is drastically
insufficient. Some people, like yourself, have reported success with
changing winecfg to use OSS, so this then uses pulseaudio's OSS
wrapper (padsp).

The real way to fix this is one of two things:
1) pulseaudio should not use direct hardware access and instead access
the soundcard via hardware mixing wherever possible, or failing that
the ALSA dmix plugin, or
2) a (nearly?) fully-functional libpulse driver should be committed to Wine.

#1 is unlikely to happen because pulseaudio IS a mixer. #2 so far has
never been submitted as a implemented solution. There is a winepulse
driver floating around, but in reality it's just a hacked-up ALSA
driver and it doesn't use libpulse.

> My value to this group is usually as a canary in the mine. Because I have the latest kernel, and the latest updates, I am able to spot problems before they hit the mainstream.
> And remember, this is a developers' group, not a users group. Experimentation is grist for the mill.

Your efforts are consistently and constantly appreciated. Your name
was even mentioned in a thread about making users "happy" with Wine
("Article on wine development strategy").

None of this changes the fact that ALSA+dmix works wherever pulseaudio
with an ALSA back-end works. If this was a problem for me (i.e., if I
used Ubuntu), I would open or comment on a bug, the jist of which
would be "Ubuntu is crippled when pulseaudio is removed". pulseaudio
*should* be optional, even if it's opt-out.

I'm still interested to know what the problem with purging pulseaudio
is, as that is the current preferred solution that is given to users
who have sound issues in #winehq.



More information about the wine-devel mailing list