M. Dietrich wrote: >> 3.) pygsmd, a minimal implementation in python using a proprietary dbus >> interface (but soon freesmartphone.org). > > i developed the pygsmd for two reasens: the first was because i wanted > a usable implementation and libgsmd did not work for me for several > reasons, the second was to learn how to best interact with the modem. > i thought that if i had learned how to serve the modem right (and i > saw that that was complicated in libgsmd) i could reimplement the > behavior in c. > > when mickey jumped in i thought the bottom-half of the learning could > be also done on the top half: implementing the right api using a > standard, easy to use interprocess communication system. that's the > point where we are now. > > next thing i will do is changing the api of pygsmd to meet the > requirements as we agreed with mickey. here we can learn if the api is > right and change the code fast for better ideas on that api. > > after this learning curve i suggest to implement the pygsmd in c. i > have seen that if i enable all daemons (there are similar daemons for > gps, pwr, ppp, ...) the python process needs a notable amount of cpu > resource.
This sounds like the sanest approach in my (uninformed) opinion. And as far as I am concerned, no need to rush the pygsmd conversion to C - as long things are still not set in stone, it's much faster to redesign things in python. But for the medium term, definitely a C implemention. -- Naked