Bug in floating point multiplication

[Irmen de Jong]

On 3-7-2015 7:07, Ned Deily wrote:
> In article <559579bb$0$2921$e4fe514c at news.xs4all.nl>,
>  Irmen de Jong <irmen.NOSPAM at xs4all.nl> wrote:
>> Tested on Mac OSX 10.10.4, with a 64-bit core2duo processor. Below are all 
>> 64-bit python
>> implementations:
>> 2.6.9 (apple supplied), 2.7.6 (apple supplied), 3.4.3 (homebrew), and 
>> pypy-2.6.0
>> (homebrew). I don't have any 32 bit Python implementations on the mac.
> 
> Sure you do!
> 
> $ /usr/bin/python2.7 -c 'import sys;print(sys.maxsize)'
> 9223372036854775807
> $ arch -i386 /usr/bin/python2.7 -c 'import sys;print(sys.maxsize)'
> 2147483647
> 

Woops! Totally forgot about the multiarch thing. Thanks for reminding me.
It doesn't change the outcome of the test though, when running the little test code with
the 32-bit versions they both (2.6 and 2.7) ran to completion.

Irmen