[Numpy-discussion] toward python 2.6: mtrand, gettimeofday and mingw

Charles R Harris charlesr.harris@gmail....
Fri Nov 14 11:42:30 CST 2008


On Fri, Nov 14, 2008 at 10:13 AM, David Cournapeau <cournape@gmail.com>wrote:

> On Fri, Nov 14, 2008 at 6:39 PM, Charles R Harris
> <charlesr.harris@gmail.com> wrote:
>
> > static float npy_tanhf(
> > ...
> > #define tanhf npy_tanhf
>
> I don't like this solution so much. The right solution IMHO is to
> correctly detect the intrinsic so we don't define a function already
> available.
>

We need a function with a pointer that can called by the ufunc. I was
guessing that with /Ox optimization windows might be using a hardware
instructions where it could  and MS forgot to put a corresponding version in
the library. In any case, the current code is safe, if ugly, and if you can
get the detection set up right it won't get in the way.


>
> In the mean time, I of course realized that the windows situation is
> more complicated than expected: we need to deal with the manifest crap
> for the configuration stage of numpy because the msvcrt 9.0 is not
> available on the average windows computer...
>

Windows Nada, the new entry level version....

Chuck
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://projects.scipy.org/pipermail/numpy-discussion/attachments/20081114/ebb8947e/attachment.html 


More information about the Numpy-discussion mailing list