The best way forward seems to be to continue to fund these existing JITs. Maybe once one is ready it can replace cpython.
Also pypy not supporting the latest python 3.5 is kind of a big deal for adoption. I would rather have 3.5 support than numpy/cpython compat but I don't do scientific computing :)
Also pypy not supporting the latest python 3.5 is kind of a big deal for adoption. I would rather have 3.5 support than numpy/cpython compat but I don't do scientific computing :)