Q: Feature Wish: "%" Extension

[Tim Peters]

[Tim]
> python-needs-a-transfinite-cardinal-type-ly y'rs  - tim

[Huaiyu Zhu]
> Is there a way to bound the computational resources needed for
> such beasts?

I sure hope not!  Half the point is to use the growing popularity of Python
to force HW vendors to get off their butts and produce systems with
transfinite memory and speed <wink>.

> I would think a type of algebraic numbers would be much more useful than
> transfinite cardinals, but to do it properly perhaps requires a whole
> computational algebra package.

If you're talking "useful", I'm not your bot.

> On a related issue: How practical would the proposed rational type be
> for serious computation considering that the results of simple
> operations are likely to grow in size for both numerators and
> denominators?

All forms of computer arithmetic are rife with surprises, from bounded ints
to the constructive reals, and all points between.  It takes an expert to
use rationals effectively for "serious computation" -- but then serious
numeric computation of any kind requires an expert.  Note, though, a
commercially important counter-example:  rationals barely grow at all when
adding a gazillion "dollars and cents" kinds of inputs, as the denominator
never exceeds 100 (or falls below it either, if rationals don't bother to
reduce to lowest terms except once before output -- as recent experience
suggests is actually best in practice, most of the time -- and, yes, you
need an expert to know when skipping normalization is going to be a
time/space disaster instead).

> just-trying-to-keep-the-thread-alive-ly y'rs

Good job!  Hitler got mentioned only once <wink>.