Re: [PATCH 0/3] Long dial string support (2nd)

[Denis Kenzior <denkenz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1602 bytes --]

Hi Andras,

> Even if we don't support FDN at this point, there is still a remaining
> related issue, who we are going to dial numbers like:
> 
> 12345456p1234#1111
> 
> I think the the dial string could still be passed to the voicecall driver
> that will take care of the modem specific details, most importantly
> playing the DTMF tones and generating the tone events that would be
> propagated back to voicecall manager and to D-Bus from there.

So we looked into what it would take to implement dial strings with AT
modems.  The consensus was that passing the entire string to the driver
was a bad idea as most modems simply do not support pause characters and
have strict limitations on the number length.

The best idea we have came up with so far is to parse the string passed
to dial and separate the actual number from the dial string.  The dial
string then gets assigned to a separate property on the voice call
object (see Dial String task in the TODO).

We would then extend the call state logic to queue the dial string the
same as a DTMF once the call is active.  If you study the DTMF logic,
you will note that Andrew has recently made it into a tone queue.

Today we burst up to 8 (arbitrarily picked number) DTMF tones per driver
request.  What we could do is send a single tone at a time.  In theory
this would allow us to emit the Tone Started / Tone Stopped signals
(please see the provide feedback of sent DTMF tones task) as well.

It sounds like the ISI modems already work this way and are nicely
covered by this approach...

Regards,
-Denis