Re: [PATCH 4/8] call-forwarding: class applied is the class given by SS code

[Denis Kenzior <denkenz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2222 bytes --]

Hi Philippe,

<snip>

>>> + /*
>>> + * Some modems don't understand all classes very well, particularly
>>> + * the older models. So if the bearer class is the default, we
>>> + * just use the more commonly understood value of 7 since BEARER_SMS
>>> + * is not applicable to CallForwarding conditions according to 22.004
>>> + * Annex A
>>> + */
>>> + if (cls == BEARER_CLASS_SS_DEFAULT)
>>> + cls = BEARER_CLASS_DEFAULT;
>>> +
>>
>> Strictly speaking this is wrong. You cannot modify the cls here as it is
>> used to generate the reply in cf_ss_control_reply. Since
>> BEARER_CLASS_SS_DEFAULT != BEARER_CLASS_DEFAULT you will generate the
>> wrong result.
>>
>
> OK, but I don't understand why we intend to generate a reply including
> classes for which we don't know the status?
>
> With BEARER_CLASS_SS_DEFAULT, we are considering the classes
> DataSync/DataAsync/Fax/Sms/Voice.

For some bizarre reason 22.030 specifies that all bearers are queried in 
the case that the mmi service code is omitted.  That means SMS as well. 
  However, half the bearers are not even applicable to call forwarding.

When reporting the results we report all bearers that were queried, even 
if some of them are not applicable or not known.  This is because we do 
not want to require the calling application to have any intelligence in 
this area.  It should simply display the results as given by oFono.

>
> But by default, the query is asking the status only for
> BEARER_CLASS_DEFAULT (Data/Fax/Voice).
> So, why, don't we match strictly the reply with the query?

We do, but querying with class is broken on many modems, hence the 
comment and why we optimize this case.  If we pass in an explicit mmi 
service code then that is what will be used instead.

If your question is why we always query with the default, the answer is 
why not?  Default literally means Voice, Data, Fax.  Where Data is 
equivalent to DataAsync | DataSync in 27.007.  So in effect we get a 
full set of results and we return only the applicable set to the 
application.  Just because the GCF testing methodology is broken in this 
area does not mean that this approach is incorrect.

Regards,
-Denis