I've had this idea rattling around in the back of my brain for a while. What if a bare-metal implementation of python for the CE was faster than TI's implementation with an ARM co-processor? It's certainly possible given how horrendously slow TI-Python is.
First off, gonna try to port Micropython. If that works I'll begin rewriting heavily used routines in Assembly.
Otherwise (and with much more time) I'll just write my own implementation.
There is a high likelihood this project won't ever get off the ground but I wanted to put up a topic so perhaps I can stop procrastinating and just do it.