Re: Telephony API, new version

Hi Zoltan,

Thanks for the reply.

On 2013?04?02? 15:14, Kis, Zoltan wrote:
> Hi Hsin-Yi,
>
> On Fri, Mar 29, 2013 at 12:11 PM, Hsin-Yi Tsai <htsai@mozilla.com 
> <mailto:htsai@mozilla.com>> wrote:
>
>     The version uses MSISDN as service identifier for cellular
>     telephony. However, MSISDN isn't a *mandatory* file on SIM, how
>     would we expect to use this as service identifier?
>
>
> Thank you for this question, the concern is totally valid. In certain 
> cases the MSISDN is not on the SIM card (e.g. prepaid cards that get 
> an MSISDN at first usage). In such cases there are ways to determine 
> the MSISDN, but these involve the services (e.g. a USSD session, or 
> WAP request). The SIM card does have an ID (ICC ID), which can be 
> mapped to the MSISDN and IMSI.
>
> We have discussed this topic, but it needs further discussion since it 
> was left in a controversial state on the editors meeting between 
> simplicity and correctness, and was not finished. For instance, if the 
> MSISDN is not available, the ICC ID could be used and the 
> implementation would map it later. By the time the SIM card is usable 
> for making or receiving calls, it has to have the MSISDN.

For some SIM cards, yes, it will have the MSISDN once it is usable for 
making or receiving calls, but not all! MSISDN is optional in spec. The 
SIM card I am using now has no MSISDN stored, for example.

ICC ID seems the proper one used for the service Id.

> Apps are likely not interested to have the service exposed through the 
> API before that happens. But such assumptions should be discussed and 
> perhaps exposed in a note.
>
> However, in other cases multiple SIM cards may actually work with a 
> single MSISDN, which indeed kills the idea of having service id's 
> covered by MSISDN. So I am positive about separating the telephony 
> service identifier from the MSISDN, and have a separate property for 
> fetching the MSISDN (if available). Since that may involve telephony 
> services, I would even go further in the design to draft a simple, but 
> extensible TelephonyService object in a separate spec (I have a 
> proposal for that), but let's do that on the F2F eventually. We know 
> what needs to be done, the design has been tested, and also have a 
> proposal on the how, but let's see if we find consensus about it.
>
> For now, I suggest we remove the constraint, but don't support 
> fetching the MSISDN through the Telephony API. IMO it should be a 
> functionality exposed in the (future) separate TelephonyService API.
>

The suggestion sounds good to me and I also support having a separate 
service API (in the future, maybe ;-) ).

Best regards,
Hsinyi

-- 
Hsin-Yi Tsai ???
Mozilla Taiwan
T: +886-2-87861100 ext:312
htsai@mozilla.com