Just a question what is wrong with static member function or global function make_xxx? The idiom of constructing object and then making additional initialization is already used in Boost.Python - borrowed function for constructing handle. Also in Python code it will be written explicit what programmer is going to do. Did I missed something? Roman. -----Original Message----- From: David Abrahams [mailto:dave@boost-consulting.com] Sent: Wednesday, July 23, 2003 5:49 PM To: c++-sig@python.org Subject: [C++-sig] Injected constructors It's already possible to inject new methods into a Python class wrapper that are not in the underlying class: #include <vector> struct X { std::vector<int> v; }; int size(X& self) { return self.v.size(); } BOOST_PYTHON_MODULE(test) { class_<X>("X") .def("size", size) ; } >>> x = X() >>> x.size() 0 Currently, there's no easy way to do the same thing for constructors, without intruding on the underlying class or writing a wrapper class which repeats *all* of the base constructors and adds new ones. I am about to add just such a facility: std::auto_ptr<X> x_factory(PyObject* self, int n) { std::auto_ptr<X> x(new X); x->v.resize(n); return x; } BOOST_PYTHON_MODULE(test) { class_<X>("X") .def("size", size) .def("__init__", init_factory(x_factory)) ; } >>> x = X(30) >>> x.size() 30 init_factory creates a python callable object which calls its function argument and injects the resulting object into a holder in the self argument. I'm not in love with the "init_factory" name. Anyone have a better idea? -- Dave Abrahams Boost Consulting www.boost-consulting.com _______________________________________________ C++-sig mailing list C++-sig@python.org http://mail.python.org/mailman/listinfo/c++-sig