[C++-sig] Re: Boost.Python scope

David Abrahams dave at boost-consulting.com
Mon Nov 10 01:12:48 CET 2003


"Ralf W. Grosse-Kunstleve" <rwgk at yahoo.com> writes:

> --- Nicolas Fleury <nidoizo at yahoo.com> wrote:
>> This is very nice, since I have a way of working with a 
>> boost::python::object.  However, my question remain, should Boost.Python 
>> scope/goal be enlarged to have no need for Py... C functions, providing 
>> a full C++ API for Python/C++ interaction?
>
> You are asking the right question. Some work on this was done more than a year
> ago. Basically the idea was to write a Python script that generates C++
> wrappers for the entire Python "C" API with all reference counting handled by
> boost::python::object and boost::python::handle<>. Very unfortunately, how
> should I say, that horse was beaten dead somehow. Now we need an intrepid new
> horse to pull this off. If someone feels brave enough to step forward I'll try
> to help and dig out the work that was done before.

Yeah, it would be great if someone would pick this stuff up and flesh
it out.

Keep in mind that the existing API wrapping facilities are designed to
keep extension modules which use the Boost.Python shared library as
small as possible, so that many inline overloads of some of the
function templates just end up converting their arguments and
dispatching to an implementation function in the shared library which
handles boost::python::object instances.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com





More information about the Cplusplus-sig mailing list