[Numpy-discussion] vectorize problem

Arnd Baecker arnd.baecker at web.de
Thu Apr 27 03:06:17 CDT 2006


On Wed, 26 Apr 2006, Travis Oliphant wrote:

[...]

> It is just a simple change.   Scalars are supposed to be supported.
> They aren't only as a side-effect of the switch to not return
> object-scalars.    I did not update the vectorize code to handle the
> scalar return value from the object ufunc (which is now no-longer an
> object-scalar with the methods of arrays (like astype) but is instead
> the underlying object).
>
> I'll add a check.

Works perfect now - many thanks!

This reminds me of some other issue when trying to
vectorize f2py-wrapped functions:
Pearu suggested a fix in terms of a more general way to determine the
number of arguments of a callable Python object,
http://www.scipy.net/pipermail/scipy-user/2006-April/007617.html

However, it seems that this has fallen through the cracks
(and I don't see how to incorporate it into numpy.vectorize...)

Is this another simple one? ;-)

Many thanks,

Arnd




More information about the Numpy-discussion mailing list