As a test, I have been trying to overload numpy’s “all” function and method to take an optional argument ‘axis’.
I have succesfully been able to overload the np.all function to achieve this:
@overload(np.all)
def jit_np_all(a,axis=None):
…
Calling the np.all function with one argument now uses jit_np_all instead of the original np_all function in numba/np/arraymath.py
np.all(a)
Calling the np.all function with two arguments also uses jit_np_all instead of the original np_all function in numba/np/arraymath.py
np.all(a, axis=1)
however, I can’t find a solution that works for overloading the corresponding “all” method on types.Array.
Running the above snippets (as a Jupyter NoteBook), I get the following errors:
Is there a way to use ‘@overload_method(types.Array, “all”)’ without causing np_all to give a typing error?