SWIG compatability with Python 2.0 final?

[Alex Martelli]

"Johann Hibschman" <johann at physics.berkeley.edu> wrote in message
news:mtn1g0zq4m.fsf at astron.berkeley.edu...
> cherokee30114  writes:
>
> > Does anyone have a comment on the compatability of SWIG with the Python
> > 2.0 final release? The documentation on SWIG's web site seems a little
> > dated and I am about to undertake a massive project integrating Python
> > into my companies C++ engine. (I can't wait to see management's faces)
>
> So far, it's worked for me.  I'm using python 2.0b2 and SWIG 1.3a1 as
> we speak, for numerics.  I don't pretend to have exhaustively tested it,
> but I'm a data point.

I haven't tried SWIG with the Python 2 final release either, but I also
had no problem using it for simple things with Python 2 betas.

For 'massive' integration of C++ stuff, though, I would recommend
either CXX (mature, massive) or py_cpp (newish, lean-and-mean).

These take specific advantage of C++ features and know the typical
C++'ish things one would like to do (e.g., expose a C++ class to
Python as a Python class, so that Python code can inherit from it,
override some virtual methods, etc... I'm not sure if CXX can do it
all, including implicit "callbacks" from C++ to Python when C++
code calls a virtual method that has been Python-overridden, but
that's a py_cpp strength).  I know there are other specific tools for
C++/Python integration (the py_cpp page mentions a few), but CXX
and py_cpp (and SWIG) are the ones I've "used in anger" and can
recommend.


Alex