[Pyrex] special floats generate incorrect C code when used as compile-time constants

[Stefan Behnel]

Hi,

Chris Mellon wrote:
> On Tue, Apr 22, 2008 at 6:13 PM, Greg Ewing <greg.ewing at canterbury.ac.nz> wrote:
>> Stefan Behnel wrote:
>>  >   DEF INFINITY = float('inf')
>>  >   print INFINITY
>>  >
>>  > results in:
>>  >
>>  >   foo.c: In function 'init2foo':
>>  >   foo.c:158: error: 'inf' undeclared (first use in this function)

This was actually due to a bug report by Jacques Frechet:

https://bugs.launchpad.net/cython/+bug/218377


>>  I'm not sure whether to do anything about this for Pyrex.
>>  As I understand it, Python doesn't promise anything about
>>  representation of inf and nan, so it's platform dependent.
>>  I'm inclined to think that using an appropriate constant
>>  from your platform's math.h is the right thing to do.
> 
> As of 2.6, float('inf') and float('nan') will be guaranteed to work
> (instead of punting to libcs atof which is what it does now), so
> supporting it in Pyrex is forward-looking at least  ;)

In Cython, a float "nan" or "inf" constant will now be translated into the C
equivalents from math.h (NAN/INFINITY). Do you happen to know if that's
compatible with what Python 2.6 does?

Stefan