[openal] A 5.1 32-bit source kills a stereo 16-bit one on OpenAL Soft

Leszek Godlewski lg at inequation.org
Tue Oct 7 02:14:50 EDT 2014


I do keep checking for errors - a habit picked up with OpenGL development.
;) There are no errors.

The high number of buffers is my "fault" since I create one for each Vorbis
audio packet, queue it and once it's played out, unqueue it and destroy it.
There is no pooling, it was simpler to integrate Ryan Gordon's TheoraPlay
this way.

Then we're in luck, because it's a Linux port. :) I'll contact you with the
details off-list.


On 7 Oct 2014 08:08, "Chris Robinson" <chris.kcat at gmail.com> wrote:

> On 10/06/2014 05:53 AM, Leszek Godlewski wrote:
>> Hey,
>> So I just checked – both sources keep incrementing their offsets and
>> rotating their buffers. There's nothing suspicious I can see in their
>> states. [...]
> The next thing to check is if you're getting any AL errors (from
> alGetError). Since you say it works when both streams have the same format,
> but one goes silent if they're different, I'd check if maybe the stereo
> stream is getting confused from the other having a different format.
>  AL_BUFFERS_PROCESSED is 0 because the custom mixer uses byte-tight
>> quadruple buffering and I just caught it in the middle of swapping.
> What do you mean by 'byte-tight'? I notice too the 5.1 stream has 148
> buffers queued with 106 processed, which is quite a lot (though that's not
> inherently a problem, it just looks a bit weird).
>  Or is anyone willing to have a look at the case in point in exchange for a
>> Steam key maybe? :)
> I could, although my main development system is Linux. If it's a Windows
> app, I'd have to run it under Wine to get useful info, which may or may not
> work.
> _______________________________________________
> openal mailing list
> openal at openal.org
> http://openal.org/mailman/listinfo/openal
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openal.org/pipermail/openal/attachments/20141007/730b96cf/attachment.html>

More information about the openal mailing list