On Wed, Aug 27, 2008 at 9:44 AM, Pierre GM <[EMAIL PROTECTED]> wrote:
> On Tuesday 26 August 2008 20:41:44 Charles R Harris wrote:
>
>> I've been thinking of reverting the changes rather than waiting for David
>> to get back.
>
> As mentioned in a previous email, that fails on x86_64 AMD as well, same
> place. Commenting out `trunc` in umathmodule.c.src does the trick.

Yes, it is not arch specific, but runtime specific. trunc is not
available from math.h without adding some flags at compilation with
glibc and gcc, but is with mac os x C runtime. I did not expect it
would have been different from ceil and other rouding methods, I
should have checked.

Since the change is reverted and we are in rc, I won't change it back now.

cheers,

David
_______________________________________________
Numpy-discussion mailing list
[email protected]
http://projects.scipy.org/mailman/listinfo/numpy-discussion

Reply via email to