<div dir="ltr"><div class="gmail_default" style="font-family:arial,helvetica,sans-serif">I though to keep changes confined to only one plugin from the beginning, but adding a new interface to MMBroadbandModemClass does make sense to me. </div></div><div class="gmail_extra"> <div class="gmail_quote">On 2 December 2015 at 19:40, Aleksander Morgado <<a href="mailto:aleksander@aleksander.es" target="_blank">aleksander@aleksander.es</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On Wed, Dec 2, 2015 at 5:50 PM, Carlo Lobrano <<a href="mailto:c.lobrano@gmail.com">c.lobrano@gmail.com</a>> wrote: > I am trying to solve a problem on HE910 Telit modem (I also opened bug 93135 > about this), which does not support NULL items for AT+CPMS command, so > commands like > > AT+CPMS="","ME" > > return with error. My idea is basically to provides always MEM1 parameter > when AT+CPMS is issued, using the current value set on the modem if not > provided in other ways. > > As suggested by Aleksander on bugtrace, the right way to do it would be to > call +CPMS? "somewhere" and parse the reply. > I'd like to keep those changes in telit plugin, but I wouldn't re-implement > all messaging interface for this and it is not clear to me how access to > "current_sms_mem1_storage" parameter in MMBroadbandModemPrivate. Is it even > possible? Well, nothing is impossible, but I wouldn't try to do it that way. The current_sms_mem1_storage variable is part of the MMBroadbandModem logic, it doesn't apply out of it, it's only used as part of the logic doing the storage lock/unlock before R/W. Instead, what I would do, is to provide in MMBroadbandModemClass a new async virtual method to "load_current_sms_mem_storage()" (or something like that). The logic in MMBroadbandModem which checks which are the supported storages could then have a new step to query which is the current storage (which we only need to do once), all this in the MMBroadbandModem class. This class would then provide an implementation of the virtual method itself, not in the Telit plugin, as the change shouldn't affect other implementations and actually kind of makes sense. If we see that the change affects other modems, we can later just move the implementation of the virtual method to the Telit plugin and leave the generic MMBroadbandModem with the virtual method pointers to NULL so that other plugins don't use that logic. What do you think? -- Aleksander <a href="https://aleksander.es" rel="noreferrer" target="_blank">https://aleksander.es</a> </blockquote></div> </div>