I have been looking at using either scheme, ML, or python for building a
speech processing environment and a 3d graphics system. After two months
of comparing the languages, python seems the clear choice. However, both
of the other languages have reasonably efficient compiliers to either C,
or native machine code. This makes it possible to prototype in the
interpreted language, and then produce fast executables.
Before I spend too much time trying to write a compiler for python, I'm
looking for the wisdom of the net.
Is there something inherent in the python system that makes this task
intractable?
Has anybody else tried building such a system (with any degree of sucess
or failure)?
Have parts of the system been written (ie. type-reconstruction for
python)?
Thanks for any suggestions.
-- Jim Hugunin hugunin@mit.edu