[Pyro-developers] Wrapping C and C++ code

dblank at cs.brynmawr.edu dblank at cs.brynmawr.edu
Sat Aug 16 13:30:35 EDT 2003


Good question! Here's my thinking:

We already have to support SWIG because at least one other package that we
depend on requires it (Aria). Aria, in fact, needs the latest and greatest
SWIG. I'm hesitant to add yet another dependency (like Pyrex or Boost), unless
we get a great benefit because there is always the risk that work will stop on
it. For example, I see that Boost is not quite ready for Python2.3. What if it
never was ready?

SWIG also makes our code available to other languages besides Python. 

However, if someone wrote (or ported) a project in/to one of these and showed
the benefit, we would probably adopt it in some fashion. Or if these projects
stay around, show their stability, and gain popularity, we might switch. I
think Boost has shown this stability.

Managing a long-term open-source project is filled with questions like this,
and ones that we must continually keep asking lest we get stuck depending on
something that is no longer around. 


"William K. Volkman" <pyro at netshark.com> said:

> I've noted that most of the wrapping that seems to be used is
> based on SWIG.  Has any thought been given to using boost.python
> for the C++ or pyrex for the C interfaces?  Just curious. The
> SWIG interfaces are a quick initial implementation however suffer
> from weakness when it comes to python reference counting or
> acting like first class python objects.
> References:
> http://www.cosc.canterbury.ac.nz/~greg/python/Pyrex/
> http://www.boost.org/libs/python/doc/index.html
> Regards,
> William
> _______________________________________________
> Pyro-developers mailing list
> Pyro-developers at emergent.brynmawr.edu
> http://emergent.brynmawr.edu/mailman/listinfo/pyro-developers


More information about the Pyro-developers mailing list