[pulseaudio-discuss] [PATCH] echo-cancel: Drop the correct amount of samples when skipping.

Tanu Kaskinen tanu.kaskinen at digia.com
Wed Apr 25 01:59:25 PDT 2012


On Wed, 2012-04-25 at 10:45 +0200, Peter Meerwald wrote:
> Hello Tanu,
> 
> > ---
> >  src/modules/echo-cancel/module-echo-cancel.c |    2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> I am wondering how you came up with these echo-cancel related patches?
> are there concrete issues you are fixing, is it code reading/review?
> 
> is there some regression testing in place? the EC module is big and 
> non-trivial
> 
> what is the situation where this problem triggers?
> 
> I am not complaining about a specific patch you submitted, just concerned 
> about the testability of the EC stuff

As part of reviewing a patch for adding assertions to
pa_memblockq_push() calls where safe, I'm investigating whether it's
safe to assert that the pa_memblockq_push() calls in module-echo-cancel
always succeed. During that investigation I've noticed some unrelated
problems in the echo-cancel code, resulting in these patches. I have not
tested the patches (testing them would probably be rather tricky). I did
ask Arun about this sample dropping patch before preparing it, though,
and it looked like a bug to him also.

Asking for testing is good, but sometimes testing just too hard to do,
and leaving a likely bug unfixed doesn't sound like a good idea.

-- 
Tanu



More information about the pulseaudio-discuss mailing list