On 07/12/2010 09:47 PM, fons@email-addr-hidden wrote:
> On Mon, Jul 12, 2010 at 09:30:05PM +0100, Rui Nuno Capela wrote:
>
>> problem is that input buffers are always read before any output signal
>> is produced, while on the same process period cycle. iow, you're
>> consuming what hasn't been produced yet. that's why you'll probably get
>> silence, at best.
>
> This should still work, you should get the output from the
> previous cycle. Unless
>
> 1. you clear outputs before reading inputs, and
> 2. Jack is using the same buffer for the feedback input
> and the output that connects to it.
>
yep, all outputs are zeroed at each cycle. i suspect some potential
garbage could be injected in the process otherwise...
> (2) is an Jack bug wich I reported at least a year ago.
> Don't know if it has been fixed. You can work around it
> by connecting a second (silent) output to the feedback
> input - this stops Jack sharing the buffer. Or insert
> a dummy client as you suggested.
>
i think that's the "zero-copy" optimization that jack pursues when ports
are single connected?
byee
-- rncbc aka Rui Nuno Capela rncbc@email-addr-hidden _______________________________________________ Linux-audio-dev mailing list Linux-audio-dev@email-addr-hidden http://lists.linuxaudio.org/listinfo/linux-audio-devReceived on Tue Jul 13 04:15:02 2010
This archive was generated by hypermail 2.1.8 : Tue Jul 13 2010 - 04:15:02 EEST