Hi I get an exception when numpy.float32 are passed to c++ functions that are expecting a float. I have provided a trivial example #include <boost/python/module.hpp>

#include <boost/python/def.hpp> double sq(double x){ return x*x;} BOOST_PYTHON_MODULE(sq) { boost::python::def("sq",sq); }

When i run it with : from sq import sq import numpy as np

print sq(3.) #fine

print sq(np.float64(3.)); #fine print sq(np.float32(3.)); #dosn't work I get: 9.0

9.0 Traceback (most recent call last): File "sq.py", line 6, in <module> print sq(np.float32(3.)); #dosn't work Boost.Python.ArgumentError: Python argument types in sq.sq(numpy.float32) did not match C++ signature: sq(double)

I know these are different types and I can just recast the input to float64, but I assumed that a float32->float64 conversion would be automatic in this case. Is this the way it is supposed to work? Is there a way to make an explicit conversion happen? Thanks, Kevin Meagher