AT%RAI=2 and AT+CFUN=0 AT+CFUN=1 cycle

[Achim Kraus]

With mfw 2.0.2, AT%RAI supports now also "2 – Activate with unsolicited notifications."

That's verfy useful, but I miss some details in the documentation about the usage of that 2.

If I set it to 2,

AT%RAI=2

and read it back

AT%RAI?

%RAI: 1
OK

I get surprisingly a 1 instead of 2 as expected by me.

If that is intended, I would expected to read something explicit in the docu about that the 2 is read as 1 and that it is not possible the check, if the notifications are enabled.

Further, and for me more important, does a AT%RAI=2 survive a AT+CFUN=0 AT+CFUN=1 cycle? Or must it be reset to 2 after AT+CFUN=0 before AT+CFUN=1 ? (Here it would help to be clear, if the read result is erroneous or if it's just not possible to check, if the notifications are enabled.)

In my opinion, having that details in the docu will help to use %RAI proper. Including the Nordic Team, which works on the NCS, may need to know, which notifications must be re-enabled in a AT+CFUN=0 AT+CFUN=1 cycle (see NRF_MODEM_LIB_ON_CFUN handler). 

I would even got that far, the knowing, which notifications in general will survive a AT+CFUN=0 AT+CFUN=1 cycle and which need to be re-enabled will help a lot. So maybe adding that behavior detail in general will be very welcome.

[Achim Kraus]

What may help in such cases would be to add a "custom hook/callback" for cfun.c/enable_notifications. It could be done in the NRF_MODEM_LIB_ON_CFUN/ cfun=0 callback, but in "enable_notifications" the usage may be more straight. 

[Michal]

Hello,

I am looking into it, I hope I will have something during the next week.

Best regards,

Michal

[Achim Kraus]

Thanks! 

There is no hurry from my side.

I found some time ago a similar issue with the SMS notification and so I started to consider, that if the modem at-cmd docu is explicit about sustainability over an power off/on cylcle, then it may be clearer.

Adding two new codes  LTE_LC_FUNC_MODE_BEFORE_NORMAL and LTE_LC_FUNC_MODE_BEFORE_ACTIVATE_LTE may be an idea for a future support of custom "enable_notifications".

[Michal]

Thank you for understanding, just wanted to let you know that I have relayed this internally.

Best regards,

Michal

[Michal]

Okay, so in general the notification subscriptions are not stored, but for the ones enabled by the LTE LC library, the configuration should be stored on the application side. For %RAI it is currently not.

Achim Kraus said:

What may help in such cases would be to add a "custom hook/callback" for cfun.c/enable_notifications. It could be done in the NRF_MODEM_LIB_ON_CFUN/ cfun=0 callback, but in "enable_notifications" the usage may be more straight. 

Yes, you are right, a solution for that is to enable all the notifications needed in that callback.

Best regards,

Michal

[Achim Kraus]

Michal said:

in general the notification subscriptions are not stored, but for the ones enabled by the LTE LC library, the configuration should be stored on the application side.

Then it's a bug, isn't it?

[Michal]

Yes, to my understanding we are going to have a fix in a future version of the LTE LC library.

Thank you very much for reporting that.

Best regards,

Michal

[Achim Kraus]

Great!