Following up on a thread that I started about Agents/Queue and acknowledging calls before bridging them... Greg Boehnlein said that he was putting his efforts into ICD. I downloaded and installed ICD, and I can get simple queue and agent stuff working fine, and see that this new design is much cleaner and more powerful. That said, in the sample conf files, the "acknowledge_call" field is labeled as "TBD", so it doesn't appear to be implemented yet. A quick scan of the c files shows it being parsed in at least one place (or so I think), but I am also not getting the debug output on the CLI that seems to be in there, so I'm either putting the keyword in the wrong place (I have it in the agent definition context) or that part of the code doesn't get hit. Anyway, the real point of this post is to point out that I am marginally surprised that there is close to zero traffic on this list regarding ICD, and I don't know if that's because no one uses it, no one has any problems with it (including wanting to get the new stuff working), or I'm just on the wrong list (I am not currently subscribed to -dev, but would head over there if this is an active topic on that list). If the authors of ICD are on this list, and prefer a private email dialogue, that would work as well, as I'm willing to be a serious tester of the app.
On Tue, 30 Aug 2005, Hadar Pedhazur wrote:> Following up on a thread that I started about Agents/Queue and > acknowledging calls before bridging them... > > Greg Boehnlein said that he was putting his efforts into ICD. > > I downloaded and installed ICD, and I can get simple queue and agent > stuff working fine, and see that this new design is much cleaner and > more powerful.We use ICD to handle a complex queuing situation. This involved a new distribution algorithm and hooks to the Asterisk management interface. ICD suffers quite a bit from over engineering. The complexity comes mainly from the same analysis that makes Asterisk overly comples - signalling is handled in parallell thread. There are some locking issues and races we have not been able to track down. For our use we no longer see any hangs, but from time to time agents enter an inconsistent state. Both Asterisk and the ICD subsystem would have been easier to implement and debug had they chosen a single-threaded model for the control flow and left the threads to handle the payload (voice etc).> Anyway, the real point of this post is to point out that I am marginally > surprised that there is close to zero traffic on this list regarding > ICD, and I don't know if that's because no one uses it, no one has any > problems with it (including wanting to get the new stuff working), or > I'm just on the wrong list (I am not currently subscribed to -dev, but > would head over there if this is an active topic on that list).ICD has its own mailinglist at Icd-user@lists.sourceforge.net. There is close to zero traffic there as well. I think the authors read it though. Peter