Re: RE: Prothon on CPython intrepreter? [PROTHON]

Mukhsein Johari <[email protected]> Wed, 11 Aug 2004 08:49:30 +0100 (BST)
Newsgroups gmane.comp.lang.prothon.user
Message-ID <[email protected]>
 --- Mark Hahn <[email protected]> wrote: 
> 
> In case I didn't make it clear in the previous
> message, I'm talking about
> implementing a C-based Prothon interpreter
> (CProthon) by heavily modifying
> the C-based Python interpreter (CPython).  This

Mark, could you give some reasons as to *why* you want
to do this? IMHO, the cool ideas you were thinking
about (stackless etc) wrt. the prothon interpreter are
much more interesting.

> 
> The Prothon language itself would not be compatible
> with the Python language
> so any part of the library written in Python would
> need to be manually
> ported.  There would be a utility to textually
> replace Python constructs
> with Prothon constructs but this would only
> accomplish 90% of the porting.

This could still be done *without* utilizing the
python interpreter. Textual replacement and manual
greps...we don't need the python interpreter for that,
do we?

Considering that we *still* have to port the vast
python libs anyway... what are hoping to gain? I'm
thinking that as more people use prothon, many will
port their favourite python modules anyway.

> So this will not be exactly the same as Prothon on
> .Net where multiple
> languages can use the library with 100%
> compatibility, but it will be quite
> close, since the most interesting libraries are C
> extensions and the Python
> to Prothon porting will be easy.
> 

Again, I don't understand why you need to use the
python interpreter to achieve this.

Right. So here are my not so humble opinions:

1. You can achieve something similar to .net via
mozilla's XPCOM. It is cross-platform, and components
can be written in any XPCOM-capable language so your
potential for reusing existing libs is there. It is
smaller than .net. It uses native components where
necessary or prudent to do so.

2. Write a small and fast interpreter (like you
planned to do before you got seduced by the .net) make
it XPCOM-capable so you can use it in gecko for
webpage scripting (prothon-DOM) as well as 'local'
gecko-based apps (prothon-XUL). This will also take
care of your GUI capability. (need to think about
sandboxes, though)

3. Take a look at mozilla-firefox style Extensions
"download, install and usage" mechanisms.

4. You mentioned interest in xml and internet libs,
well...mozilla has this and more. (CSS, DOM, RDF) The
mozilla community is quite large and backed by an
aggressive open source non-profit (Mozilla
Foundation).

5. Implement the interesting ideas you discussed wrt.
garbage collection.

6. Design and implement psyco-like JIT to get "C-like"
performance.

I *did* say "not so humble"! Thing is, I think prothon
with the above would be very compelling, not to
mention bold and exciting. Imagine using a single
language for client-side scripting as well as
server-side programming (I'm thinking twisted here) as
well as conventional GUI apps! And banking on w3c
standards to boot. The only language fitting that bill
right now is the crummy javascript. :-/

________________________________________________________________________
Yahoo! Messenger - Communicate instantly..."Ping" 
your friends today! Download Messenger Now 
http://uk.messenger.yahoo.com/download/index.html