[asterisk-dev] Re: [svn-commits] murf: trunk r41283 - /trunk/main/pbx.c

[Steven Critchfield]

On Wed, 2006-08-30 at 12:40 -0500, Jason Parker wrote:
> ----- Steven Critchfield <critch at basesys.com> wrote:
> > On Wed, 2006-08-30 at 10:57 -0500, John Lange wrote:
> > > On Wed, 2006-08-30 at 10:31 -0500, Steven Critchfield wrote:
> > > > > Already the existing system has some limitations; for example it
> > can not
> > > > > match a variable number of digits. I had a friend who was trying
> > to
> > > > > implement an ISDN lookup for library books (ISDNs are variable
> > length)
> > > > > but had problems because of the limitations on the matching.
> > > > 
> > > > I think you meant ISBN. And variable number of digits is
> > supported. You
> > > > just need to know how to determine when it is shorter or longer
> > and
> > > > write your dialplan as such. 
> > > 
> > > LOL, ya, ISBN. Apparently I've exceeded the maximum acronym storage
> > > limit of my brain.
> > > 
> > > > Think about it, in dialplans, most people have short internal
> > extensions
> > > > that can be dialed, and longer PSTN local and even longer PSTN LD
> > > > numbers that can all be handled just fine. 
> > > 
> > > I'll ask for the specific example. I think it had to do with
> > terminating
> > > with a "#". So:
> > > 
> > > exten => _X.#,1,....
> > > 
> > > Does not work. Everything after the "." is ignored and you are
> > forced to
> > > rely on the digit timeout for asterisk to start doing anything. This
> > is
> > > exceptionally annoying in this case because the timeouts are long
> > to
> > > allow people to enter a difficult number.
> > 
> > That _may_ be considered a bug. I hadn't thought of useing the #
> > terminater in the dialplan.
> > 
> > Of course it could have been implemented in a loop. loose idea below,
> > not sure it would work exactly as written.
> > 
> > [isbn-input]
> > exten => s,1,Playback instructions.
> > exten => _X,1,do something to set/add this collected digit into a
> > var.
> > exten => _X,2,Goto(isbn-input,noop)
> > 
> > exten => noop,1,noop()
> > 
> > exten => #,1,validate the number
> > exten => #,2,goto(somewhere else)
> > 
> > Defiantely jumping through a little hoop, but otherwise not bad. I
> > would
> > like to see the example you mentioned be able to work though.
> >
> 
> The problem with using # in a dialplan, is that many SIP (and probably
> other) devices silently eat #.  # is often the "send the call with the
> previously dialed digits" button.

I could see that being a problem if you where trying to dial all this at
the begining, but should sip eat any DTMF at all once the call is
established? Would make many IVRs unhappy then.


-- 
Steven Critchfield <critch at basesys.com>