Re: vendor models and options

[Denis Kenzior <denkenz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1599 bytes --]

Hi Giacinto,

> yes, CGAUTH can be queried, but there is no fallback command in case it 
> fails, so it has to be pre-defined based on the model used.

You might have to show me more specifics.  I don't see why we couldn't 
query +CGAUTH=? in all cases and if it isn't supported, fall back to 
vendor specific commands?

> No query commands would not return the name of the parameters, which in 
> this case is the only difference between username and password.

That's too bad.  Some vendors define a different length for username / 
password, so you could guess the order by which lengths are returned. 
Others use weird strings with password or username in the '=?' response.

However, since you likely want to use a vendor specific gprs-context 
driver anyway, this really shouldn't be that big of a deal.

> 
> I am introducing a specific gprs-context for some models, using atmodem 
> for others, and mbim and qmi for other models.
> I try to stick to gprs (not gprs-context) for most of them, because this 
> is pretty much the same for most of them. For some models that have a 
> lot of differences we are going to use specific atoms.
> We use the same code for the lte atom as well, unless there is a way to 
> factorize it.

It is generally quite easy to share things between AT command based 
drivers.  See drivers/atmodem/atutil.[ch].

Also, you might want to start by submitting a small subset of changes 
you have pending as a trial run to get comfortable with the overall 
development process, patch reviews, resubmissions, etc.

Regards,
-Denis