Re: vendor models and options

[Denis Kenzior <denkenz@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1774 bytes --]

Hi Giacinto,

On 09/07/2018 03:09 AM, Giacinto Cifelli wrote:
> Dear Denis, all,
> 
> while preparing the Gemalto driver, I see that there is a potential 
> shortcoming with the vendor structure in the atmodem. I can set 
> OFONO_VENDOR_GEMALTO_model, but this could be a very long list for all 
> options to maintain. I see the tendency also for other manufacturers.
> Would it be ok if I convert the vendor integer in a structure with model 
> and flags?
> This would make the current code more compact and clearer.

Feel free to suggest something more concrete illustrating how your 
proposal would result in more compact code.  It seems to me that it 
would be quite disruptive, but I always try to have an open mind :)

But of course there are other 'creative' ways we can utilize to handle 
this as well.

> 
> The alternative of cloning the atmodem is less tempting, because then it 
> needs constant monitoring of the atmodem to porting the features in it, 
> or conversely missing features in this general driver. One example is 
> for the indicators +CGREG/+CEREG/+C5GREG that we are adding in atmodem 
> (gprs.c), instead of the unique +CGREG that doesn't work anymore for LTE 
> (in the 27.007). If I add it in Gemalto driver, it will be missing in 
> the atmodem.

Aren't these commands now standardized?  Why do you need sub-model 
information to handle this?

Either way, it is quite hard to give you any suggestions without seeing 
concretely what you're trying to do.  Perhaps you want to submit what 
you have (or some subset that illustrates the issues you're having) to 
the mailing list (even if it isn't completely ready, just tag it as 
'RFC') and then I can give you more concrete feedback.

Regards,
-Denis