Wet Dream--python to native compiler

Travis N. Vaught travis at enthought.com
Sat Dec 14 16:06:50 EST 2002


hwlgw at hotmail.com (Will Stuyvesant) wrote in message news:<cb035744.0212120320.1fda37c0 at posting.google.com>...
> As someone else mentioned you need a lot of grad students or similarly
> capable people.  And then...what platform do you want to target?  Many
> universities now force linux on their scientific personnel so you are
> not going to see a free open source compiler for the windows platform.
> 
> And is it even the right approach to compile to assembly?  History has
> seen the development:
> binary ->
> assembly ->
> C
> 
> This suggests we should compile to C instead of assembly. 
> Unfortunately the py2C project has failed because of lack of man- and
> brainpower.  And there is the issue of deciding what libraries we
> assume etc.
> 
> BTW, there *is* a free C compiler for every platform.
> 

One approach is to get a former graduate student (now Ph.D.) and a
National Labs guru to collaborate on some crazy byte code magic --
even though it's unsanctioned and not financially viable.  You could
then produce weave (http://www.scipy.org/site_content/weave) and, if
it is able to take 'the next step' you could have an automagic way to
get native C speed out of python (after the initial compilation).

See: weave.accelerate

This is important, elegant stuff if you ask me.

Travis



More information about the Python-list mailing list