[Asterisk-Users] Re: app_conference and AGI

[Jean-Hugues ROBERT]

At 15:31 07/07/2005 +0000, Tony Mountifield wrote:
>In article <6.2.1.2.0.20050707114929.036f5990 at pop.mail.yahoo.com>,
>Jean-Hugues ROBERT <jean_hugues_robert at yahoo.com> wrote:
> > But... what is wrong with MEETME_AGI_BACKGROUND ? (besides the fact
> > that unfortunately it does not work for SIP channels due to the mixing
> > not being done in the zaptel driver but app_meetme itself, sort of, AFAIK).
>
>It's the other way round. The mixing is always done in the zaptel driver.
>For non-Zap channels, MeetMe creates a Zap pseudo channel, and in its main
>loop it copies frames of audio in both directions between the non-Zap
>channel and the associated pseudo channel. The Zaptel driver mixes the
>audio in the pseudo channels and the real Zaptel channels.
>
>When using MEETME_AGI_BACKGROUND, the main loop that does the pseudo
>channel copying is not invoked, so only the hardware channels get mixed in
>the Zaptel driver.
>
>Cheers
>Tony

Thanks for the clarification Tony.

Any idea on how to make it so that MEETME_AGI_BACKGROUND would work on
SIP channels (well, I suspect the issue is there with IAX too or any
VoIP channel for that matter...) ?

Maybe there could be a thread that would do what the main loop does.

But... there might be an issue if the two threads (the one dealing with
AGI and the "main loop" one) both try to read the frames...

If this is not possible, then maybe the copying should occur earlier, before
frame is delivered to the AGI ? This may require an additional data
member in the channel structure. Well... this is kind of beyond my
current needs/knowledge.

OTOH, isn't recording done in a distinct thread ? If so, then the same
kind of solution might be feasible.

Thanks again for the clarification.

Yours,

     JeanHuguesRobert

-------------------------------------------------------------------------
Web:  http://hdl.handle.net/1030.37/1.1
Phone: +33 (0) 4 92 27 74 17