If you're a hobbyist buying a microcontroller kit that needs an expensive development kit then you're most likely buying the wrong thing. At one point I considered cataloguing the different microcontroller solutions that are friendly to the average Free Software host or which provide a Free Software toolchain, but with so many decent, reasonably priced products out there, it would have been a huge task.
For example, all of the Arduino solutions (based on both AVR and ARM) use development tools that are Free Software, and you can even ignore the Processing-based IDE and use gcc if you want. Browsing around various online stores yields products using other microcontrollers that also seem to be supported by Free Software and which don't cost hundreds of dollars just to get started.
I'll agree with you that the Raspberry Pi is probably more convenient than something like Arduino for developing and deploying embedded software, but there are plenty of people who will justifiably argue that you aren't really doing microcontroller or truly embedded work at that point but are instead doing the kind of interfacing that used to happen a lot more back in the early days of the microcomputer era. Not that this is a bad thing if that's what you really want to do, of course.
Posted Jan 5, 2013 23:32 UTC (Sat) by dlang (✭ supporter ✭, #313)
[Link]
Given that the Pi folks are explicitly looking back at the early microcomputer era (the C-64 in particular) as a guideline, the fact that the Raspberry Pi looks very much like that means they succeed :-)
I think there's also a lot of value in having a computer that can easily interface with hardware devices rather than just a microcontroller.
Doing simple things is simple on both devices, but then when you want to start doing UI and network type things, having a complete OS and related toolsets available is much better than having to try and rebuild the base functionality on a microcontroller.
Now, if you need thousands of devices, then the cost difference becomes a driving factor, but for one-off projects, learning, and experimentation, going the full computer route is a good approach.