[SciPy-user] Segfault in scipy.sparse module

[David Cournapeau]

Hi,

    I got a segfault with scipy.svn on the sparse module. I think I saw 
a similar report on a Mailing List, but cannot find it. My configuration:

    - winxp
    - atlas 3.8.0 compiled by myself with cygwin (I also tried with pure 
blas/lapack from netlib)
    - mingw32 compilers
    - both numpy and scipy are today checkout.

The segfault always happens at the add_sub_test (TestBSR). This seems at 
least os/compiler specific, since it does not happen on linux with gcc. 
Maybe some problem between g++ and the msvc runtime ?

cheers,

David