As maintainer of the gpsd package I started to wonder what the opinion of the gpsd developers about gypsy is - especially after reading that "Gypsy was designed to fix the numerous design flaws found in GPSD". Well, here is their answer, which I support fully.
I'm blogging about it as I think it is important to know about this details if you plan to do something useful with a gps device, for example on mobile phones like the OpenMoko Freerunner. gpsd works very well on embedded devices, and if there's a feature missing just contact the developers. They're very responsive and like to implement useful features, if they can get a spec to do so (and even better, the hardware to test it). So I'm really wondering if it wouldn't make sense to replace the minimal NMEA/UBX implementation in the FSO Framework Daemon by gpsd, even if that means to compile code instead of writing Python code, which is much faster for rapid prototyping, of course.