The np.size() Function Fails

[RoyiAvital]

The Bug

It seems Numba support the ndarray property of an array (See numpy.ndarray.size) yet fails with the function np.size().

Code for Producing the Bug

The following works:

import numpy as np
from numba import njit

@njit
def SumArrayWorks( vX : np.ndarray ) -> float:
  arrSum = 0.0
  for ii in range(vX.size):
    arrSum += vX[ii]

  return arrSum

vX = np.array([1.0, 2.0, 3.0, 4.0])
SumArrayWorks(vX) #<! Works!

While this will fail:

import numpy as np
from numba import njit

@njit
def SumArrayFails( vX : np.ndarray ) -> float:
  arrSum = 0.0
  for ii in range(np.size(vX)):
    arrSum += vX[ii]

  return arrSum

vX = np.array([1.0, 2.0, 3.0, 4.0])
SumArrayFails(vX) #<! Fails!

The error is:

TypingError: Failed in nopython mode pipeline (step: nopython frontend)
Use of unsupported NumPy function 'numpy.size' or unsupported use of the function.

System Information

1. Numba: 0.59.0.
2. Python: 3.12.
3. NumPy: 1.26.4.
4. OS: Windows 11 Pro 23H2.

[sklam]

np.size() is so rarely used that we completely missed it for all these years.

[SDanielDev]

Hi, would a simple solution be to just call np.size() outside of the njit function? I am new to this, but here is my attempt to use np.size() over .size:

import numpy as np
from numba import njit

@njit
def SumArrayWorks_NpSize( vX : np.ndarray ) -> float:
  arrSum = 0.0
  for ii in range(vX_size):
    arrSum += vX[ii]

  return arrSum

vX = np.array([1.0, 2.0, 3.0, 4.0])
vX_size = np.size(vX)

SumArrayWorks_NpSize(vX) #<! Works!```

[RoyiAvital]

@SDanielDev , The concept here is about having a native Numba support, not bypassing it.
Specifically, it is better to avoid using global variables. A simpler and better solution would be using vX.size instead of np.size() which works as shown above.