[SciPy-dev] ***[Possible UCE]*** Re: ***[Possible UCE]*** Re: Another segfault

[Travis Oliphant]

Nils Wagner wrote:

>>
>>Nils
>>    
>>

I was able to reproduce the problem.   Several issues.  Basically, the 
error-handling in certain segments of _superlu module was bad (things 
being called that shouldn't be called and a couple of missing checks for 
NULL). 

Then the error that was being caught badly was a result of a corner-case 
incompatibility in the PyArray_CopyFromObject function which I've fixed 
in NumPy now.

So, now I'm not getting segfaults (but still errors that I'm trying to 
figure out).

Thanks very much for the tests.

-Travis