On Tue, Nov 13, 2007 at 11:18:49AM +0000, Ondrej Certik wrote:
> 
> > Besides, if we do, it would be better to try Pyrex or Cython, not C++,
> > since they allow easy incremental conversion, are easier to use, and
> > play much better with Python itself.
> 
> Yeah. BTW, on of the guys who is doing Cython (Robert) is sitting
> right next to me now. Cython is cool. But still you need to modify the
> original python source, the pypy's approach
> is better imho - but pypy is unfortunately not ready for production.
> Cython guys are really making sure it just works. And also it's very
> clean - the result of of Cython is one nice *.c file, which you just
> compile and you get one *.so file and that's it. Much better than
> SWIG, that creates a bloated and slow interface. BTW, if you use
> Debian, I created a preliminary deb package here:

I've used Pyrex and had positive experience in how well it integrates
with Python from one side, and C from the other side.

Yes, Pyrex (or Cython) is a compromise and pypy is an extreme case. But
compromises are good!

(And even in pypy, I've heard there is RPython which is sutable for JIT.
Haven't tried it, so just trying to mention it for completness)

> > P.S. premature optimization is the root of all evil (Donald Knuth)
> >      http://en.wikipedia.org/wiki/Optimization_(computer_science)
> 
> Very true. I think SymPy will be ready for it, when we find a robust,
> simple, general and a fast way how to do assumptions.

Hmm... to early? maybe...

I think the first thing we should do, is to merge in steps, with
reviewing, sympycore ideas and patches into main repository.

It's the design that is hard, coding is relatively simple. And we should
be confident that our design is mature before doing optimizations.


> On the other hand, I think it could be useful to have a simple C++
> implementation of the very few basic SymPy classes. Too see how fast
> it is. And then compare it with doing the some in Cython for example.

Do it, if it has motivations for you.

I myself was many times in a situation when I optimized something, then
redesigned, then saying 'I was so stupid to spend so much time wasting.
What a fool!'

It seems the story repeats again and again.


> We may even code in it C. C is a nice language

Yes, C is nice
+1

> when you know what you want to do and how.

This is important.


-- 
    Всего хорошего, Кирилл.
    http://landau.phys.spbu.ru/~kirr/aiv/

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To post to this group, send email to sympy@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/sympy?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to