RPython is a translation and support framework for producing implementations of dynamic languages, emphasizing a clean separation between language specification and implementation aspects.
By separating concerns in this way, our implementation of Python - and other dynamic languages - is able to automatically generate a Just-in-Time compiler for any dynamic language. It also allows a mix-and-match approach to implementation decisions, including many that have historically been outside of a user’s control, such as target platform, memory and threading models, garbage collection strategies, and optimizations applied, including whether or not to have a JIT in the first place.
This is an experimental release of a randomly chosen, untested version of RPython. Packaging issues are likely, feedback is welcome.