Usability problem: Cannot access .shape member of numpy.ndarray as tuple

On typed variable of type numpy.ndarray, the .shape member does not work

cdef numpy.ndarray var = numpy.zeros((3,))
print var.shape

gives the cython compiler error “Cannot convert ‘numpy.npy_intp *’ to Python object”

Comment by dagss (mailing list):

This is mostly a question of semantics. shape is defined to be a C integer array for speed of var.shape[0]. How can we define a generic rule in Cython which makes both possible?

Should the behaviour of “var.shape” e.g. depend on the requested return value? This means overloading on return value which is a non-trivial language change, though perhaps appropriate here.

Thoughts welcome!

Migrated from http://trac.cython.org/ticket/302

[robertwb]

bfroehle commented

Not very satisfying, but of course the following will work:

cdef np.ndarray var = np.zeros((3,))
print (<object> var).shape

[scoder]

BTW, doesn’t arr.shape[:3] (for a 3D array) work, and provide a nicer work-around? That’s basically what Cython would have to do internally then.