[Numpy-discussion] Current SVN segfaults under VC7

Travis Oliphant oliphant.travis at ieee.org
Wed Mar 22 19:07:08 CST 2006

Tim Hochberg wrote:
> OK, now that I've got things working, I can report a real segfault 
> under VC7. The following snippet of code:
>    from numpy import *
>    a = arange(100.)
>    b = array(1+1j)
>    (a+1) ** (b+3)

I can reproduce it on Linux.   It looks like it's related to the power 
optimization stuff that was recently checked in.   power(a,b+3) does not 
crash but

 >>> a**(b+3)

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1076248704 (LWP 1545)]
0x404162df in array_power_is_scalar (o2=0x8294200, exp=0xbfffec68)
    at arrayobject.c:2656
2656    arrayobject.c: No such file or directory.
        in arrayobject.c


More information about the Numpy-discussion mailing list