Garth Coghlan
2009-05-05 01:59:47 UTC
Hi,
I've been wrapping some code with PyUblas, and appreciating its
usefulness. Today, though, I have discovered that I cannot successfully
pass a numpy array with dtype='uint8' through to a C++ function with a
numpy_matrix<unsigned char> argument. It complains that the signatures
don't match. The same also happens with char/int8. I can do int32s and
floats fine.
Possibly not related, but I noticed that in numpy.hpp the following line
was commented out:
inline NPY_TYPES get_typenum(npy_ubyte) { return NPY_UBYTE; }
So I uncommented that and commented the bool line instead. But that
hasn't helped.
Any ideas?
Thanks,
Garth
--
I've been wrapping some code with PyUblas, and appreciating its
usefulness. Today, though, I have discovered that I cannot successfully
pass a numpy array with dtype='uint8' through to a C++ function with a
numpy_matrix<unsigned char> argument. It complains that the signatures
don't match. The same also happens with char/int8. I can do int32s and
floats fine.
Possibly not related, but I noticed that in numpy.hpp the following line
was commented out:
inline NPY_TYPES get_typenum(npy_ubyte) { return NPY_UBYTE; }
So I uncommented that and commented the bool line instead. But that
hasn't helped.
Any ideas?
Thanks,
Garth
--