[asterisk-dev] Unified Asterisk Interface

Tzafrir Cohen tzafrir.cohen at xorcom.com
Tue Dec 11 11:29:24 CST 2007


On Tue, Dec 11, 2007 at 05:56:31PM +0100, David Roden wrote:
> On Tuesday 11 December 2007 17:24:30 Steve Edwards wrote:
> 
> > > Right. Currently I have to add three different AGI calls to an Asterisk
> > > dialplan for my application to work. This is something I would like to
> > > get rid of, or cut down at least to the required minimum--which would be
> > > a single line.
> > Don't dump on AGIs too much :)
> > Even a wimpy 1.6gHz Celeron can execute 100 AGIs per second.
> 
> You're not understanding me. Performance is the least of my worries. When I 
> want to place an outbound call from my Java application I have to do all 
> this:
> 
> 1. Setup an extension (say, "dialout") that executes AGI(agi://host/out).
> 2. Setup an extension ("control") that executes AGI(agi://host/control).

Why aren't those already set up in advance in the dialplan?

Isn't that what contexts are for?

Also: what is the "control" AGI for? Shouldn't you use AMI for
controling other channels?

> 3. Place an outbound call via AMI to Local/dialout at context that jumps to 
> extension "control" on connect.
> 4. Accept the incoming connection on dialout.
> 5. Dial the number I really want to reach and keep the connection.
> 6. Accept the incoming connection on control and keep it waiting.
> 7. Trace various renames of the involved local channels so that I know which 
> channel is the real outbound channel which has to be watched for events.
> 8. On hangup, notify the two waiting threads so that all channels can be taken 
> down.
> 
> (Okay, actually 1 and 2 only have to be done once.)

(After reload?)

I get the impression that you actually try not to touch the user's 
dialplan at all costs.

-- 
               Tzafrir Cohen
icq#16849755              jabber:tzafrir.cohen at xorcom.com
+972-50-7952406           mailto:tzafrir.cohen at xorcom.com
http://www.xorcom.com  iax:guest at local.xorcom.com/tzafrir



More information about the asterisk-dev mailing list