[asterisk-dev] [Code Review] Design review: allow remote AGI to effectively use EAGI

[Tilghman Lesher]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/524/#review1548
-----------------------------------------------------------


One additional thought I had.  Currently, this requires that the client AGI set up a socket and listen on it for Asterisk to connect to, which may not be easy for some languages.  Perhaps it would be helpful to additionally have a passive mode, whereby the command does not directly open a connection, but merely waits on a particular port, which is specified back in the AGI protocol exchange.  The remote process would then be responsible for connecting to that port before audio would flow.

- Tilghman


On 2010-02-22 00:08:44, Tilghman Lesher wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviewboard.asterisk.org/r/524/
> -----------------------------------------------------------
> 
> (Updated 2010-02-22 00:08:44)
> 
> 
> Review request for Asterisk Developers.
> 
> 
> Summary
> -------
> 
> Various people have asked on the mailing list whether it was possible to get the audio stream from a remote AGI script.  This is usually not possible, because by nature, a remote AGI is connected across a single socket, and audio data typically needs to run across a separate socket, to avoid confusing the command syntax engine.  However, we could certainly open another socket on demand from the AGI script, once it has started, by allowing the AGI script first to listen on a network port for connections, then specifying that hostname and port across the AGI interface to Asterisk.
> 
> 
> This addresses bug 16879.
>     https://issues.asterisk.org/view.php?id=16879
> 
> 
> Diffs
> -----
> 
>   /trunk/res/res_agi.c 247785 
> 
> Diff: https://reviewboard.asterisk.org/r/524/diff
> 
> 
> Testing
> -------
> 
> Compile testing only.  As I said, this is at design phase, and changes may be in order to the "open audio" command before we get to testing stage.
> 
> 
> Thanks,
> 
> Tilghman
> 
>