In article <[EMAIL PROTECTED]>, Hendrik van Rooyen <[EMAIL PROTECTED]> wrote: > >"Stef Mientki" <[EMAIL PROTECTED]> wrote: > >> I use Python as a replacement for MatLab, >> and intend to use it as replacement for Delphi, AutoIt, PHP, VB. >> And I'ld love to use it as a replacement for micro controller programming. > >If you have a little sub - 64k micro this is not really practical. > >What we have done is to define a limited virtual machine with >a simple "assembler" language and used Python to write the >"compiler" for it. > >Then you have to write the interpreter for the virtual machine >in the native micro's code - Either using C or assembler, and >you are away from the hardware, into a useful layer of abstraction. > >... well sort of - in such a small thing, you are never really far from >the hardware, as that is what its supposed to be controlling - but >what it definitely buys you is that your apps written in your special >language become portable between disparate processors like say >an ARM at the upper end and an 8031/8051 at the lower... > >Such "interpreters" are surprisingly little code, and they work >well - faster than what one would expect. - often outperforming >C code that does pass by value... > >- Hendrik >
Amen! That is, while it is indeed easy to construct examples where Python computes a numerical result only a hundredth as fast as the corresponding C code, what I find far more interesting to discuss are the (surprisingly frequent) cases where Python and litte languages result in solutions which are *faster* at run-time, along with being more maintainable and quick to develop. Yet more chatter on a role for Python with specialized hardware ap- pears in <URL: http://www.unixreview.com/documents/s=9133/ur0404e/ >. -- http://mail.python.org/mailman/listinfo/python-list
