[asterisk-dev] [Code Review] New dialplan function to allow for inheritance of audiohooks

[Mark Michelson]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviewboard.digium.com/r/102/
-----------------------------------------------------------

(Updated 2008-12-18 17:15:30.748561)


Review request for Asterisk Developers.


Changes
-------

I realized after viewing the second version of the diff that the TODO comment would fit better in audiohook.h instead of func_audiohookinherit.c, since it will be the audiohook code which will need to be modified to allow for multiple audiohooks from the same source to be moved.


Summary
-------

It has been a problem for a while now that MixMonitor does not continue to record a conversation if the channel on which the MixMonitor is attached is masqueraded (the most common case being a transfer). MixMonitor is not the only audiohook source that does not propagate to new channels during a masquerade, and so I made the dialplan function generic enough to allow for any audiohook source to be transferred properly.

This patch introduces a new dialplan function called AUDIOHOOK_INHERIT. The basic mechanism is that if one declares an audiohook from a specific source as being inheritable, then the audiohook will move to the new channel structure during a masquerade. I have included example usage of the function in the XML documentation in the file.

One point to double check in this is the behavior change I made to ast_do_masquerade. I changed the datastore processing a bit by calling the fixup function only on the datastores on the clone channel and doing so before moving them onto the original channel. As far as I can tell, this actually makes more sense than how it was done previously, but I'd like to know if this could cause troubles.

I know of one limitation right now, which is that only one audiohook of a given source may be moved to another channel during a masquerade. So if someone had things set up so that there were two MixMonitors on a channel, and that channel got masqueraded into a new one, only the first MixMonitor audiohook found while traversing the list of audiohooks would be moved. At this point, the effort required to accommodate moving multiple audiohooks greatly exceeds the need to implement it, so I am holding back unless it is deemed necessary.


This addresses bug 13538.
    http://bugs.digium.com/view.php?id=13538


Diffs (updated)
-----

  /trunk/funcs/func_audiohookinherit.c PRE-CREATION 
  /trunk/include/asterisk/audiohook.h 165810 
  /trunk/main/audiohook.c 165810 
  /trunk/main/channel.c 165810 

Diff: http://reviewboard.digium.com/r/102/diff


Testing
-------

I have tested by using MixMonitor on a channel which gets masqueraded into a new channel as a result of transferring. I have tried both attended and blind transfers, both with the channel as the caller and callee. The result was that the recording continued across the transfer.

I also did some testing where I set a ChanSpy audiohook on a channel which is masqueraded. I specifically tested this to be sure that the datastore fixup done in app_chanspy would not interfere with the datastore fixup being done in func_audihookinherit. Things worked as expected.


Thanks,

Mark