[Numpy-discussion] Scalar math module is ready for testing

Ted Horst ted.horst at earthlink.net
Sun Apr 30 16:50:08 EDT 2006


Here is an issue I am having with scalarmath:

 >>> import numpy
 >>> numpy.__version__
'0.9.7.2462'
 >>> import numpy.core.scalarmath
 >>> a = numpy.array([1], 'h')
 >>> 1*a
array([1], dtype=int16)
 >>> 1*a[0]
Traceback (most recent call last):
   File "<stdin>", line 1, in ?
TypeError: unsupported operand type(s) for *: 'int' and 'int16scalar'

This happens because PyArray_CanCastSafely returns false for casting  
from int to short. alter_scalars(int) fixes this, but I have lots of  
non-numpy code that I don't want to behave differently.

Ted

On Apr 28, 2006, at 02:12, Travis Oliphant wrote:

> The scalar math module is complete and ready to be tested.  It  
> should speed up code that relies heavily on scalar arithmetic by by- 
> passing the ufunc machinery.





More information about the NumPy-Discussion mailing list