[SciPy-dev] [SciPy-user] segfault in scipy.random.standard_normal

Travis Oliphant oliphant.travis at ieee.org
Fri Dec 2 19:54:00 CST 2005


Robert Kern wrote:

>Chris Fonnesbeck wrote:
>  
>
>>Something is squirrely with the size argument in standard_normal:
>>
>>In [5]: random.standard_normal()
>>Out[5]: 0.31727273308342824
>>
>>In [6]: random.standard_normal(2)
>>Segmentation fault
>>
>>I'm using a recent svn build from scipy_core.
>>    
>>

I've been mucking with the C-API lately and whenever that happens 
strange errors can occur unless you rebuild everything.   

The reason for the C-API changes and the big ones coming has been 
getting the numarray records module working better than I had previously 
envisioned.  These changes will be for the better and should be the last 
ones for awhile.  I try to change the version number whenever I alter 
the C-API, so new version number means new C-API.

After the records module is ported (and improved IMHO), the C-API 
changes will settle down.   Then, the only changes I can forsee in the 
future are additions to ease the porting of numarray extensions like 
nd_image and PyTables.

I'd love it if somebody started thinking about a good way to ease the 
numarray transition.


-Travis





More information about the Scipy-dev mailing list