ANN: Experimental Number Types (Integer, Rational, Floats)

[M.-A. Lemburg]

"Delaney, Timothy" wrote:
> 
> > > > The normal way to create a Rational is to write Rational(2, 3).
> > > > The next release will also have a string parser for rational
> > > > numbers (in the format "2/3" and probably "12 2/3" too).
> > >                                              ^^^^^^
> > >                                              ^^^^^^
> > > Just a note:
> > > This one is cultural and could generate lots of confusion.
> > > The only way I can parse it is 12*2/3 (I'm French).
> > > No French book (I know of) uses this notation: 38/3 or 12+2/3
> > > is used instead. It's as if you wrote the complex number
> > > 3+4j as "3 4j" (look: "3 4j + 4 5j * 1 7j", hmmm...).
> >
> > The notation is quite common in financial business and
> > mathematics. I agree that it can be confusing, but since the
> > parser will only accept a single rational in the string, I
> > believe that at least when using Rational() this behaviour
> > can be accepted. The situation is different for literals,
> > of course...
> 
> Of course, we should probably aim for a notation which is consistent *now*.
> 
> 12 2/3 - ambguous across cultures.
> 
> 38/3   - not ambiguous, but falls into the problem of integer truncation.
> Nevertheless, I would prefer this notation, and deprecate integer division
> notation in favour of 'div' (since this would be a major incompatible
> language change anyway, might as well make a new keyword at the same time
> ;).

If rationals make it into the core some day, 38/3 might come out
as rational, so there would not be any integer truncation. The truncation
would happen when converting the rational to an integer.
 
> 12+2/3 - not ambiguous and valid if the above is valid (integer + rational),
> I would probably want the compiler to start rolling literals together at
> this point... otherwise there would be 2 runtime calculations going on
> (division to form rational, then addition of integer).

As Alex already mentioned, complex numbers work this way already
and so would rationals (this works today). What I was talking about
is conversion of strings to rationals -- this is a somewhat different
task than recognizing literals.

Still, you have valid points here and I think Moshe should consider
these in his PEP (that's why I put him on CC -- hi Moshe ;-)

-- 
Marc-Andre Lemburg
______________________________________________________________________
Company & Consulting:                           http://www.egenix.com/
Python Pages:                           http://www.lemburg.com/python/