wishlist Python3K: subclass ints

[Moshe Zadka]

On 8 Jun 2000, Aahz Maruch wrote:

> In article <393FD0C7.D70479A2 at visionart.com>,
> Pete Shinners  <pete at visionart.com> wrote:
> >
> > - Also gone are all the nit-picking type checking routines.
> >   instead of first checking if a variable is a "class" type,
> >   then checking if the object is a certain class, all variables
> >   hold objects of the same "rank". the best move would be to
> >   have "int" and "float" be subclassed from a base class like
> >   "realnumber".
> 
> Close, but no cigar.  Start with the base class "number".  Then
> everything else follows: int, float, rational, complex.

Actually, in the number case specifically, I'd go for the
no-hierarchy-have-a-good-model-optimize-behind-the-covers strategy:

an int is a rational is a complex, any number might be exact or inexact.
Inexact is implemented internally as floating-point, of course, but it's
quite a good warning to users.

Note that a really crazy extension might have non-rational exact real
numbers (e.g., constructive reals), and it's a difficult question of how
to integrate it. Lucky for us, Python is a New Jersey language, so we can
just ignore the question.

--
Moshe Zadka <moshez at math.huji.ac.il>
http://www.oreilly.com/news/prescod_0300.html
http://www.linux.org.il -- we put the penguin in .com