Builtin Float Epsilon? (was: Re: Does python suck or I am just stupid? )

Mike Meyer mwm at mired.org
Mon Feb 24 17:24:48 CET 2003


Irmen de Jong <irmen at -NOSPAM-REMOVE-THIS-xs4all.nl> writes:

> I think that what I meant was: are there languages that don't use
> "IEEE (?) floating point" numbers at all, but use some other representation
> form that doesn't suffer from the problems we're talking about?

IEEE is just one specific floating point format. There are language
implementations that have floating point numbers (by which I mean they
keep an exponent and a fraction, and the represented number is the
fraction times some known base raised to the exponent) that aren't
IEEE. I don't know of a language that specifies the floating format
that doesn't use IEEE. I don't know of any floating point
implementation that doesn't have the problems we are discussing,
though I can imagine one.

There are languages that support things *other* than floating point
for representing non-integer values: fixed point, rationals and BCD
all come to mind. As stated elsewhere, they have *other* problems -
most notably lack of a hardware implementation. Rationals also have
space problems, and can't represent irrationals at all.

        <mike
-- 
Mike Meyer <mwm at mired.org>			http://www.mired.org/home/mwm/
Independent WWW/Perforce/FreeBSD/Unix consultant, email for more information.




More information about the Python-list mailing list