Woof wrote:
> On Mon, 27 Apr 2009 08:55:35 -0400, Paul Mossman 
> <[email protected]> wrote:
> 
> > 1. Completing a Consultative Transfer to a voicemail box results in 
> > the prompt being re-started for the Transferee.  A very nice touch.
> 
> It does?  That's totally unintentional.  Might be a nice 
> touch, but it's due to a bug, not a feature!
>
> >  My complaint though is that the original prompt plays for 
> a couple of 
> > seconds before being re-started.
> 
> Well, the whole thing is a bug.  
> 
> Woof's rule: "Audio and timers from the IVR shouldn't stop or 
> restart due to external changes."  
> 
> There was a whole to-do on the SIPPING list years ago about 
> HOLD and VoiceXML based IVR systems.  I argued the IVR should 
> ignore HOLD and just keep on playing (as it would with any 
> telco interface other than SIP).  I lost, and the VoiceXML 
> SIP interworking drafts claims HOLD should freeze all audio 
> and suspend all timers, except for certain "total call" 
> timers, oh, and interdigttimers if they had pressed one digit 
> but not yet completed another.  See why this is starting to 
> smell, and why I disagree?

Yes, that smells.  

But consider softening your rule:  The IVR is not required to stop or
restart audio and timers due to external changes, though it may.

No need to do it for hold/unhold.  But if it accepts an INVITE with
Replaces, then re-starting the current audio prompt in that call would
be a nice touch.

 
> That said, I really don't know what FS does.  Try it with the 
> AA, that would have identical behavior to the new VM.

See XX-5643: Consultative Transfer to AA fails - call is dropped


> > 2. Bulk deleting multiple voicemails through the user 
> portal results 
> > in a rapid succession of message-summary NOTIFYs, one for 
> each deleted 
> > message.  Each NOTIFY is <100ms apart, in violation of RFC 3842:
> > Notifiers SHOULD NOT generate NOTIFY requests more frequently than 
> > once per second.  Ideally there would be a single NOTIFY send on a 
> > bulk delete.
> 
> At the moment, we aren't touching the user portal or the MWI 
> server 

OK.


> (although Scott wants us to scrap the MWI server and 
> build it into the new VM).  

Makes sense.  Is that the plan?


> I was aware of this issue, so if 
> we do rework the MWI server that'll be part of it.  We had 
> this discussion on the list previously.  I said I'd consider 
> delaying sending the NOTIFY by a second each time a new 
> request came in.  Thus the NOTIFY wouldn't be sent until 1 
> second after the last change.  Scott said a second was too 
> slow, and the little blinky lights wouldn't be right, and 
> people would complain.

The challenge is that a bulk delete in the User portal results in a
series of individual delete requests?  It would be easier to handle a
single delete request that indexes multiple messages.  

That way the MWI server would be able to send a "0 messages" NOTIFY
promptly, as long as it hasn't sent another within the last second.  

But yes, I understand now that XX-4753 does not include a new MWI
server.


-Paul
[email protected]

_______________________________________________
sipx-dev mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-dev
Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-dev

Reply via email to