[asterisk-dev] [Code Review] Clean up func_odbc's dummy channel usage.

[Matthew Nicholson]

> On 2009-06-23 21:49:19, Tilghman Lesher wrote:
> > While I kind of understand why you're proposing this change, I'd really prefer that you did it a different way.  The method that I used was primarily to help avoid bugs by ensuring that the channel exists at all times throughout the execution.  If you wanted to eliminate the autoservice and the setting of variables on a bogus channel, I think that would be a better way to go, but the bogus channel existence I believe should be kept for the duration.  The only benefit I see there is a temporary easing of memory usage, which isn't that great, anyway.  I think the consistency of having a channel, which helps avoid future bugs is a better way to go.

The reason for this change is mostly CEL related.  In the CEL branch, there is a new function to allocate a dummy channel that basically only has a varshead structure allocated.  This dummy channel does not support all of the operations that may happen on a channel, such as autoservices and datastores.  A normal channel cannot be used here in the CEL branch because all normal channels generate CEL events and a "bogus" normal channel would generate bogus events.  Having a dummy channel throughout the function will not work within those constraints.


> On 2009-06-23 21:49:19, Tilghman Lesher wrote:
> > /trunk/funcs/func_odbc.c, lines 243-248
> > <http://reviewboard.digium.com/r/290/diff/1/?file=5717#file5717line243>
> >
> >     I'm not really sold on this change.  Add an else clause, if you want to handle the error, instead of changing how this works.

This change was made to match the way other parts of asterisk handle channel allocation failure.


- Matthew


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviewboard.digium.com/r/290/#review874
-----------------------------------------------------------


On 2009-06-23 17:54:19, Matthew Nicholson wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviewboard.digium.com/r/290/
> -----------------------------------------------------------
> 
> (Updated 2009-06-23 17:54:19)
> 
> 
> Review request for Asterisk Developers and Tilghman Lesher.
> 
> 
> Summary
> -------
> 
> This patch cleans up func_odbc's use of dummy channels a bit.  This is accomplished by releasing the dummy channel after we no longer need it and skipping sections of code that are pointless to execute without a channel.
> 
> 
> Diffs
> -----
> 
>   /trunk/funcs/func_odbc.c 202792 
> 
> Diff: http://reviewboard.digium.com/r/290/diff
> 
> 
> Testing
> -------
> 
> None.
> 
> 
> Thanks,
> 
> Matthew
> 
>