On Tue, Dec 20, 2011 at 10:31, Marc POINOT <Marc.Poinot@onera.fr> wrote:

Hi all,

I've just changed to cython and old numpy module with a raw C API. The C module init is removed, and I've put the import_array in the 'pure-cython' part of the module init. Usual tutorial examples have these lines:

import  numpy as NPY cimport numpy as NPY

NPY.import_array()

But this fails (first numpy API call core dump) in my module, if I put back the import_array() in the C part of the module everything turns ok again. Now if I remove again this C API import_array and I write:

import  numpy as NPY cimport numpy as CNPY

CNPY.import_array()

all is ok. Do I miss something? It sounds good to me to have to separate scopes but I wonder if it's a common practice with potential side-effects or if it is the right way to use numpy with cython.

Yes, it is the right way to do it. Cython can *mostly* tell from context whether to resolve NPY.<whatever> from either the C side or the Python side, but sometimes it's ambiguous. It's more often ambiguous to the human reader, too, so I try to be explicit about it. I don't really know why the tutorials do it the confusing way. -- Robert Kern "I have come to believe that the whole world is an enigma, a harmless enigma that is made terrible by our own mad attempt to interpret it as though it had an underlying truth."   -- Umberto Eco