Re: [PATCH] voicecall: allow P or , within the dial string

[Denis Kenzior <denkenz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 4322 bytes --]

Hi Pekka,

> 2009/10/8 Denis Kenzior <denkenz@gmail.com>:
> >> "DTMF control digit separator" is used as pause according to GSM 02.07.
> >> According to GSM 11.11 it is stored as BCD code 1100 in the SIM card.
> >> Nokia modems use 'p' to represent pause. AT commands use ',' to
> >> represent pause.
> >
> > Strictly speaking AT modems use 'C' to represent the pause.  Can you
> > point me to any spec that actually says ',' and 'p' are used the way you
> > describe?
>
> C?

Unless I'm missing something, the DTMF 'C' digit refers exactly to what you're 
trying to achieve with 'p' and ','

Check TS 11.11 Table 12.  'C' is mentioned as the DTMF Control digit separator 
from TS 02.07.

Also, quoting 27.007 for valid input to VTS:  "A single ASCII character in the 
set 0 9, #,*,A D. This is interpreted as a single ACSII character whose 
duration is set by the +VTD command."

It sounds like 'p' is entirely an MMI convention, not anything defined by a 
standard...

>
> RFC 4967 uses 'X' to wait for call completion, and 'P' for pause.
>

oFono does not (yet) care about SIP.

> Nokia modems use 'p' to represent DMTF control digit separator. GSM
> 02.07 specifies how the DTMF control digit separator is used when
> making calls. The dtmf digits in the dial string are sent after the
> call gets connected. The extra separator characters represent extra
> delays.

Yes, however 02.07 never mentions 'p' or ','.  Quoting: "The way in which the 
separator is entered and display in the ME, is left to the individual 
manufacturer's MMI."

Since 'C' is defined by the standard, that is what oFono core should use.  If 
Nokia modems prefer 'p' over C, simply convert these in the specific driver.

> The dial string stored  in the SIM card along with a phone number
> cannot contain DTMF digits A-D. However, the other dialstrings do not
> share the same limitation and they can can contain full set of DTMF
> digits, 0-9, *, #, A, B, C, and D. The special bcd coded digits (dtmf
> control digit separator and wildcard) from the SIM card must be mapped
> outside this set.

SIM storage is a nightmare.  The problem is that on the SIM the following 
characters are valid: 0-9, 'A', 'B', 'C', 'D'.  'F' is used as an endmark 
(e.g. empty) character.  'E' has been deprecated and not used for anything.  
Much of the confusion stems from the fact that 'A' maps to '*' and 'B' maps to 
'#'.  Whenever you store anything on the SIM, you have to convert '*' and '#' 
back to 'A' and 'B'.  oFono does this for you when writing to / from the SIM.

The EFadn, EFsdn, etc can definitely contain the full set above for the SSC / 
Dialing Number fields.  However, the 'C' and 'D' characters cannot actually be 
sent to the network.

In fact 'D' is particularly evil since it requires special handling in the UI.  
Hence 'D' should be stripped from the valid DTMF set that oFono supports.

'C' is used as a separator in its first occurrence, but also as a pause 
character for all subsequent occurrences.

>
> Nokia modems use 'p' and '?' for that purpose (extended bcd code
> 0b1100 => 'p', extended bcd code 0b1101 => '?').  I think G1 uses ','
> instead of 'p'. Either one is fine.

Yes, you're talking about the 'C' (1100) and 'D' (1101') DTMF characters, as 
above.  I have no problem handling 'P', 'p', ',' and 'X' or whatever at the D-
Bus API level.  However, what gets sent down to the modem must be according to 
some standard.  Doing otherwise leads to chaos.

> An why to include pause in the tones string in the first place? The
> current API does not provide any feedback of the sent dtmf digits to
> the application or between ofono driver and core, so the only way to
> implement pauses between digits is to include them in the dial string
> and hope the driver can do something about them.

I still wonder about this myself.  Strictly speaking all modems should support 
the 'C' digit.  In the worst case the modem driver can simulate a 3 second 
delay itself.  Whether this should be encouraged or not is questionable.

'D' should be never sent to oFono in the first place, it is entirely an MMI 
issue.

However, I agree that the current situation is less than ideal.  Ideas / 
suggestions welcome.

Regards,
-Denis