proposed language change to int/int==float (was: PEP0238 lament)

Tim Hochberg tim.hochberg at ieee.org
Fri Jul 27 16:24:12 CEST 2001


"Guido van Rossum" <guido at python.org>:
> bokr at accessone.com (Bengt Richter) writes:
> > ISTM floor couldn't make an exact number out of an inexact one
> > unless the inexactness was known to be constrained to the halfopen
> > interval above the floor.
>
> Good point.

I guess the question becomes, does "exactness" mean just precision, or both
precision and accuracy. The result of floor is infinitely precise, but it's
accuracy is potentially suspect.

It seems that any[1] operation that converts imprecise numbers to precise
numbers is going to have this problem , so it seems best just to define
"exactness" to be "infinite precision" and live with the consequences.


-tim


[1] I suppose you could map all numbers to a constant and I suppose that
wouldn't have any accuracy problems, but it's not very interesting.





More information about the Python-list mailing list