[PYTHON MATRIX-SIG] Something broken with __mul__?

[tim@lassi.ece.uiuc.edu]

Konrad Hinsen writes:

<...>

>Just to add *why* multiplication with * fails: it doesn't
>accept an integer and a Float32 as arguments. Making either
>or both arguments float solves the problem.

I just looked and PyArray_CanCastSafely in arrayobject.c is set up
only to allow conversions from ints, longs, etc. to doubles, not to
floats. Why? I'm betting it's a typo, but I've been known to be wrong
before. If int->float is meant to be allowed, it's an easy fix:

	case PyArray_LONG:
		return (totype > PyArray_FLOAT); 

is changed to:

	case PyArray_LONG:
		return (totype >= PyArray_FLOAT); 
	

in lines 1709-10 of arrayobject.c.

>I don't remember exactly the result of the long debate about
>coercion and combination of different data types, but somehow
>I think this is not correct...

That would be my guess as well...

-- 
	-tim

+--------------------------------------------------------------------+
| Tim Hochberg               Ultrahigh Speed Digital Electronics Lab |
| tim@lassi.ece.uiuc.edu              University of Illinois         |
| http://dogbert.ece.uiuc.edu/~tim         (217) 333-6014            |
+--------------------------------------------------------------------+

=================
MATRIX-SIG  - SIG on Matrix Math for Python

send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
=================