[Numpy-discussion] I think something broke badly today in numpy/scipy...

Travis Oliphant oliphant.travis at ieee.org
Sat Jul 8 04:33:01 CDT 2006


Fernando Perez wrote:
> Hi all,
>
> I updated earlier today (about 4 hours ago) numpy/scipy SVN, and all
> of a sudden my codes broke left and right.  Backing off to
>
>   
Hey Fernando.   I think I found the problem.   It was the same problem 
causing the BFGS test to fail in SciPy.

It can be shown by looking at

sk = [0,-1.0,1.0]
print 250.0*sk[:,newaxis]

This should give something useful (not zeros). 

The problem was a seemingly harmless change in _IsContiguous that 
allowed for 0-strided arrays to be called contiguous.  This caused havoc 
with the multiplication function later which relied on an accurate 
_IsContiguous function. 

This change was in r2765 (I suspect you checked out a change with that 
in it).  The change could cause problems where-ever you use newaxis with 
scalar multiplication.

The problem should be fixed in SVN.

Very sorry...

-Travis





More information about the Numpy-discussion mailing list