[asterisk-dev] [Code Review]: app_queue: Improvements to per member 'ringinuse' (formerly 'ignorebusy') option

[jrose]

> On May 17, 2012, 12:24 p.m., Mark Michelson wrote:
> > /trunk/apps/app_queue.c, lines 7209-7215
> > <https://reviewboard.asterisk.org/r/1919/diff/1/?file=27913#file27913line7209>
> >
> >     You've got a bit of a problem here. For statically defined queues, it's possible that we will begin visiting member config settings before the ringinuse option has been set for the queue. This may result in an inaccurate setting for members of the queue.
> >     
> >     Note that this problem does not exist for realtime queues and queue members because all queue settings are applied first and then all queue member settings are applied. It works itself naturally this way since queues and queue members are separate tables.

I suppose that might be considered a problem, but it's hardly a big one and it could even be seen as useful for making the setup of members easier.  For instance,

[queuename]
;queues that should ringinuse
ringinuse=yes
member => ...
member => ...
member => ...
...
;queues that should not ringinuse
ringinuse=no
member => ...
member => ...
member => ...
...

which saves the person configuring the need to have 5 or 6 commas on every single member which has that value set.

As long as this is documented, I think most people would actually prefer it to work this way.  'Fixing' this would be moderately annoying since it would basically necessitate making the ringinuse of a member require a seperate variable indicating that it has been deliberately set and if not, then using the queue ringinuse.  If this is still not seen as ideal though, I'll make the necessary changes.


- jrose


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/1919/#review6239
-----------------------------------------------------------


On May 15, 2012, 10:59 a.m., jrose wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviewboard.asterisk.org/r/1919/
> -----------------------------------------------------------
> 
> (Updated May 15, 2012, 10:59 a.m.)
> 
> 
> Review request for Asterisk Developers, Mark Michelson, irroot, and Matt Jordan.
> 
> 
> Summary
> -------
> 
> The bug was simply the catalyst for this patch. There was another patch that actually resolved this bug...
> 
> Here's the summary:
> First, only the member setting for ringinuse is checked when determining whether to call the busy member or not.
> The default for this value is now determined by the queue's ringinuse option rather than being standalone.
> 
> additions
> Adds the ability to specify member ringinuse value as an argument
> Adds 'queue set ringinuse' CLI command for setting ringinuse for a member on the fly
> Adds AMI command QueueRingInUse for setting ringinuse for a member on the fly via AMI
> 
> changes
> 'ignorebusy' is now called 'ringinuse'
> Legacy support for old dialplans/scripts/etc will still be matched via the QUEUE_MEMBER function argument 'ignorebusy', but 'ringinuse' works as well and is the documented argument name.
> Legacy support for realtime 'ignorebusy' is in place, but at startup of the module if no instances of its use are present in the database or if a use of the 'ringinuse' field is detected, that will be used exclusively instead.
> 
> 
> This addresses bug ASTERISK-19536.
>     https://issues.asterisk.org/jira/browse/ASTERISK-19536
> 
> 
> Diffs
> -----
> 
>   /trunk/CHANGES 366295 
>   /trunk/apps/app_queue.c 366295 
>   /trunk/configs/queues.conf.sample 366295 
> 
> Diff: https://reviewboard.asterisk.org/r/1919/diff
> 
> 
> Testing
> -------
> 
> I tested a number of configurations of ringinuse for queues and members both realtime and otherwise to make sure the setting went in correctly under each case.
> 
> There is an odd quirk though for realtime in that if all the values of a field for every entry are NULL, then it will be as though that field doesn't exist at all, so if ignorebusy is present without any deliberately entered values, ringinuse will still become the defacto field name in use.  I worked at it for a while but couldn't find a workaround that didn't involve having to change every single realtime database driver.  It's not that big of a deal since if a user isn't actually using that field at all, it's unlikely they ever intended to.
> 
> 
> Thanks,
> 
> jrose
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20120517/ac7dd505/attachment-0001.htm>