[pulseaudio-discuss] [PATCH v4] Make module loopback honor requested latency

[Tanu Kaskinen]

On Wed, 2015-02-25 at 18:48 +0100, Georg Chini wrote:
> On 25.02.2015 17:51, Tanu Kaskinen wrote:
> > If there was the kind of function that you propose, would it return
> > false in this scenario? If so, that would just make module-alsa-card.c
> > kill the sink input, because it's not going to cancel the profile change
> > just because some sink input refuses to be moved.
> >
> No, it would not return false in this situation, but I when you read
> the description of the may_move_to() function it says:
> 
> 1) this function is called before the sink input is moved away

The documentation says "moved to a sink", not "moved away from a sink".
So technically it's correct, but a note should indeed be added that the
sink input may already have been detached from the old sink.

> 2) if it returns false the move will not be allowed
>
> Both is wrong. During profile switching the sink is already invalid
> when the function is called and the profile will be switched anyway,
> regardless what the function returns.

If the function returns false, moving won't be done, so I don't think
the documentation has issues here. In case of a profile switch, the sink
input will be killed. I wouldn't call that moving.

> That's why I thought maybe there should be a function for which both
> is true. But never mind - it was just a thought and I have worked around
> the issue meanwhile. But maybe the description of the may_move_to()
> function should be modified.

Good that a solution was found already!

-- 
Tanu