Implementing a scripting language in Python

[Alex Martelli]

Peter Otten wrote:

> Gabriele Farina wrote:
> 
>> I guys
>> 
>> I planned to create a new programming language to use to develop web
>> applications for my company.
>> 
>> I'd like to develop it using Python, but I don't know if python is fast
>> enaught.
> 
> The Python developers, regarded as smart over here, used C :-)
> However, there seems to be an effort under way to implement Python in
> Python, so it should be doable.

Sure.  And somebody European, like Gabriele, who's at all interested
in that, should RUN to http://codespeak.net/pypy/ and get a plane
(cheap -- I've found a trip there and back for EUR 120 + taxes, though
that specific special offer is over now) and lodging (we're doing our
best to find cheap ways for that too) to be in Amsterdam on Dec 14-21
for the fifth pypy "sprint".  With the Amsterdam effort we plan/hope
to release a big but not-too-slow cpypy.so (actually generated via
pyrex) -- not quite independent from CPython yet, of course, but still,
a complete and usable Python runtime whose sources are in Python (we
have one now, but running it in purely interpreted Python isn't exactly,
ahem, FAST:-).  The Python->C translation (with some type annotation)
that's intended to give us cpypy.so from the current Python pypy sources
is done via pypy itself, btw, with a special objectspace.  ((Plus, we
plan/hope to release cool tools for interactive display/debug of pypy
AND most particularly superior testing tools.))

I know it sounds incestuous and makes one's head spin, but if you
study all the materials available on the pypy site (including the
complete pypy sources), as is of course advisable before coming all
the way to the sprint, it really ain't all that bad -- try it.  And
then the Sprint experience is going to be VERY advisable for anybody
interested in programming language implementation of and/or with Python --
if your head doesn't explode you'll know a LOT more when you go back,


>> Someone could help me and give me some ideas?
> 
> Web applications are not rocket science, so why not adopt an existing
> language instead of burning your company's money. While we're at it, why
> not use Python - I admit I'm biased - and spend your resources on
> improving an existing web framework.

Yes, it IS nearly inconceivable that creating a new programming
language -- rather than using Python and the wealth of tools &c already
available for it -- is going to be cost/benefit-effective for a company.

Or, using Ruby, Lisp, Scheme, etc, etc, are all choices likely to be
still more cost/benefit-effective than developing a new language, even
though I agree with you that Python's likely to be best (in most, though
not all, situations of web application development -- e.g. if the firm
happens to have a Ruby guru or three already onboard, that may change
the equation).


Alex