[asterisk-dev] [Code Review] 2545: res_parking: Generate extensions when parkext is set for a parking lot. Generate extension to dial parker on parked call timeout.

[jrose]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/2545/
-----------------------------------------------------------

(Updated May 21, 2013, 10:38 p.m.)


Review request for Asterisk Developers, Matt Jordan and rmudgett.


Changes
-------

Address Matt's findings.  The registrar is handled by the config now and when reconfiguring as a preapply handler the parking extensions are cleaned up and regenerated.  If any extension fails to be added (or the extension that exists there is incompatible with the needs of a new parking lot configuration) then the configuration will be rejected and all of the extensions cleared out. If this was a reload, the previously configured parking lots will be still be alive since the new configuration was rejected, but extensions for that lot will not be available (the reload kills the old extensions prior to registering old ones and if the reason for the parking configuration failing isn't known then we can't safely restore the old extensions).


Bugs: ASTERISK-21645
    https://issues.asterisk.org/jira/browse/ASTERISK-21645


Repository: Asterisk


Description
-------

Extension generation and removal for:
Park a call to the parkext in the parking context
Retrieve parked calls from the space numbers in the parking context
Hints for ParkedCall extensions

Extension generation (but no removal yet) for:
Dial the peer which parked the call on parking timeout in the park-dial context

Extension removal for the park-dial extensions will eventually be handled when reloads are implemented. Since they can't reliably be said to belong to
the parking lot, they are registered to the base registrar for the parking system instead.

Applying bridge features in the way features.c did (as arguments to dial) is no longer necessary on account of how features are handled now.

Each parking lot has its own registrar which is 'res_parking/<parking lot name>'
The registrar used for non-lot-specific extensions is just 'res_parking'

When a parking lot is destroyed or reloaded, all of its registered extensions are flushed and rebuilt.

Note: ParkedCall extensions no longer belong to the parked user involved with them unlike how old parking works. This means that if a parking lot is reloaded and the new configuration specifies that the slot the old call was parked at no longer exists, the extension to retrieve that call will go away even though the call is still parked there. This will resolve itself when the parked call times out (provided timeout hasn't been disabled), and manually created ParkedCall extensions can still retrieve that call if the parking space option isn't provided.


Diffs (updated)
-----

  /team/group/bridge_construction/CHANGES 389247 
  /team/group/bridge_construction/include/asterisk/pbx.h 389247 
  /team/group/bridge_construction/main/pbx.c 389247 
  /team/group/bridge_construction/res/parking/parking_bridge_features.c 389247 
  /team/group/bridge_construction/res/parking/parking_controller.c 389247 
  /team/group/bridge_construction/res/parking/res_parking.h 389247 
  /team/group/bridge_construction/res/res_parking.c 389247 

Diff: https://reviewboard.asterisk.org/r/2545/diff/


Testing
-------

Checked that extensions are generated and destroyed for Park and ParkedCall as expected with multiple parking lots.
Overlaps are currently allowed. If two parking lots occupy the same space and parkext exclusive is in use, it's possible for one parking lot's extensions to be overwritten in favor of another. When parking lot extensions are overwritten by other parking lot extensions, warnings are issued.
Checked that extensions are generated and appropriately called in park-dial when a parked call times out.
Checked that the timeout is set to the parking lot's parkdialtimeout value.


Thanks,

jrose

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20130521/9379fd4c/attachment.htm>