On Jun 14, 2013, at 4:47 AM, Fernando Jiménez Moreno <[email protected]> 
wrote:

> Hi folks!
> 
> Thanks Jonas and sorry for the late reply.
> 
> On 16/05/2013, at 06:22, Jonas Sicking wrote:
> 
>> Hi Guys,
>> 
>> Spent the last few days digging my way out of email and got to this one.
>> 
>> The main problem here sounds like the fact that the user may incur
>> charges for either the incoming, the outgoing or both SMSs. To make
>> matters worse we have no idea how big charges. And to make matters yet
>> worse, I assume that once BlueVia knows the phone number, it might
>> realize that it doesn't have the appropriate business relationships in
>> place to actually charge that number?
>> 
>> So what can we do to lessen these problems?
>> 
>> One thing we could do is to check the SIM card to see if we can get
>> the phone number from that, and if we can avoid sending the MO message
>> (we'd still need the MT message). This obviously won't always help,
>> but seems like it often could. Though maybe not in our currently
>> targeted markets?
> 
> I wish we could do this, but I am afraid that it is not going to be possible 
> because:
> 
> 1. First of all, there are privacy (and probably legal) issues with allowing 
> the payment provider to get the MSISDN from the device. We've been warned 
> about this several times on TEF side. The payment provider should never get 
> the plain MSISDN. Instead, it will get a hash of the MSISDN that the carrier 
> (BlueVia in TEF case) will provide. Take a look at the flow at [1], I am 
> talking about the "Device ID" value.
> 
> 2. The SIM card usually doesn't have the MSISDN and even if it has it, we 
> cannot trust this value as the user can change it. Take a look at point 
> 4.2.27 from [2], the MSISDN field can be updatable.
> 
>> Second issue is that we should make sure to avoid international
>> charges if we can. Can the number that we are sending messages to/from
>> be made local? We should be able to detect even scenarios like the
>> user currently roaming to determine if international charges will
>> apply or not.
> 
> If I am not wrong, short codes are network specific, so for example an SMS 
> sent to an spanish short code while roaming from Germany won't work. I am not 
> sure what would be the story on the operator side in that case. I don't know 
> if the SMS will need to be sent to a regular number with the usual 
> international charges or if the SMS can be sent to a local short code. 
> Ernesto, Jorge or David in CC might know about this.
> 
> This is quite an edge case. Note that the SMS flow should ideally only happen 
> during the first payment flow as a fallback for other authentication 
> mechanisms. I agree that we still need to solve this though.
> 
>> Basically what I'm suggesting is to surface the following information
>> to the payment provider:
>> * User's current phone number (if available from SIM)
> 
> As I previously said, I am afraid that this is not an option.
> 
>> * User's current MCC/MNC (could be multiple for multi-sim devices)
>> * User's home MCC/MNC (could be multiple for multi-sim devices)
> 
> We can expose the SIM(s) MCC/MNC and a roaming flag.
> 
> I can't think about any problem in exposing this information to the payment 
> provider. However, if I am not wrong, Antonio mentioned that there are some 
> privacy concerns involved and that we probably need to notify and ask 
> permission to the user for this.
> 
>> * Ability for payment provider to send a MO message to an arbitrary
>> number (including which sim for a multi-sim device)
>> * Ability for payment provider to be notified when there's an incoming
>> message from a particular number. This notification would include the
>> message body.
> 
> This can be done :)! Probably with a few changes in the SMS implementation 
> that we'll need to discuss though.
> 
>> This way the payment provider can display UI which lets the user know
>> how many messages will be sent, and if they will be
>> international/roaming messages or not.
>> 
>> Or is this overly complex? Should we skip the MCC/MNC stuff and let
>> the payment provider simply show UI saying "you may be charged for one
>> or two SMS messages which may be international. This is a one-time
>> cost"?
> 
> For me, the problem here is not the complexity, but the possible privacy 
> issues. If notifying the user about the exposure of the MCC/MNC (with a user 
> friendly message :)) solves the privacy concerns, I am totally up to expose 
> this info to the payment provider.

The SMS authentication flow already prompts the user for a challenge and it's a 
pretty quick prompt. If we are talking about replacing the SMS challenge with a 
silent SMS that also includes a user prompt then it may not be worth our time 
:) We began discussing the silent SMS flow simply to optimize the first time 
payment experience so that it happens quicker -- that is the only reason.

Kumar

> 
> Thanks!
> 
> / Fernando
> 
> [1] https://wiki.mozilla.org/Apps/ID_and_Payments#Payments_Data_Flow_Diagram
> [2] http://www.3gpp.org/ftp/tsg_t/tsg_t/TSGT_05/Docs/PDFs/TP-99186.pdf

_______________________________________________
dev-b2g mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-b2g

Reply via email to