Work on gsmd as a GSoC project

Andy Powell openmoko at automated.it
Sun Mar 9 21:23:49 CET 2008


On Sunday 09 March 2008 14:36, Michael 'Mickey' Lauer wrote:
> Hi Sudharshan S,
>
> > I would love to work on libgsmd this summer as part of the GSoC. I am
> > aware of the work done by Mickeyl and emdete for freesmartphone.org, so
> > would like to have some comments from the list. (No one likes duplicated
> >  effort ;D).
>
> Righto.
>
> > As far as libgsmd is concerned, I would love to help around in two areas.
> > 1.) Complete the API according to the specifications and do some cleanup.
> > 2.) Implement gsm0710 specs (which freesmartphone.org does with dbus?).
> > I happened to come across some patches[1] in the bug tracker which
> > brings the mux in a lower level. Maybe it would be useful if we pull
> > that together and make it work?.
>
> The gsm situation is complicated. We soon have three things to chose from:
>
> 1.) original gsmd, which the community thinks has lots of problems.
> 2.) gsmd2 (from ixonos people), a complete reimplementation using a
> proprietary dbus interface (but soon freesmartphone.org)
> 3.) pygsmd, a minimal implementation in python using a proprietary dbus
> interface (but soon freesmartphone.org).
>
> I personally(!) think gsmd(1) is a dead-end, because of its design. It is
> already more complicated than necessary. Now that we have the
> freesmartphone.org GSM 0710 MUXer, we could simplify the design a lot.
>
> 2) may look good, but I did not have a chance to review the implementation
> yet.
>
> 3) is working pretty solid, but in Python, which may be a problem for some
> platforms.
>
> My plan a couple of months ago was to create a phone server with a backend
> and frontend so that we could play with different backend implementations
> -- eventually settling on the most stable one. (See
> http://wiki.openmoko.org/wiki/Roadmap/Telephony)
>
> Now that the freesmartphone.org OTAPI starts to stabilize, work on that
> phone server could be started. I don't know whether this could be a student
> project though.
>
> What do the others think?

Personally I think the big problem is that if this most fundamental thing 
hasn't been decided already, or at least very soon this project is going to 
get so far behind that the GTA08 will be out before it's even near 
completion. Essentially, pick the very best of those listed and stick to it. 
Both 2 and 3 seem to be using the same interface (dbus) so in theory can be 
swapped. That would tend to rule out the existing gsmd.


All IMHO...

Andy



More information about the gsmd-devel mailing list