No float limits

D-Man dsh8290 at rit.edu
Wed Jul 4 03:34:41 CEST 2001


On Tue, Jul 03, 2001 at 05:47:58PM -0600, Bjorn Pettersen wrote:
| > From: D-Man [mailto:dsh8290 at rit.edu]
| > 
| > On Tue, Jul 03, 2001 at 11:37:51PM +0200, Jürgen Hermann wrote:
| > | Hi!
| > | 
| > | I just stumbled over the fact that there is sys.maxint, but 
| > no sys.maxfloat
| > | and sys.epsfloat. In case anyone thinks this is a valid 
| > addition, I'll come
| > | up with a patch.
| > 
| > What value would sys.maxfloat have?
| > 
| > IIRC IEEE754 states that all ones (ie 0xFFFFFFFF for a single
| > precision float) means infinity, so there really isn't a "max" value
| > for a float.
| 
| Perhaps not, but there is a maximum representable number, a.k.a. DBL_MAX
| in your C library (same argument for DBL_EPSILON of course ;-)

Ok, I'm over my head now.  I'll let you float gurus fight it out :-).

-D





More information about the Python-list mailing list