[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Scheme-reports] Orthogonality in the numeric tower

Biep scripsit:

> I haven't voted because I don't grasp the implementation issues
> involved,

It's clear that even the smallest implementations such as Chibi
can support the full tower; the ones that don't do so, don't
because they have other concerns such as ease of implementation.
See <http://trac.sacrideo.us/wg/wiki/NumericTower> for a list of
which implementations do what.

The question is, what should a Scheme application programmer be
entitled to rely on having when they start to use a Scheme implementation
intended for the ordinary needs of (non-embedded) application

> but wouldn't it in principle be a Good Thing if in the Scheme numeric
> tower the various aspects were orthogonal?

In principle, yes.  However, innovation is not the point of
standardization except where it can't be avoided.  As you can see at the
NumericTower page, I have factored the tower into four orthogonal ideas:
unbounded integers, unbounded rationals, inexact numbers, and complex
(non-real) numbers.  But this was a post hoc attempt to make sense of
what already existed.

John Cowan          http://www.ccil.org/~cowan        cowan@x
Wer es in kleinen Dingen mit der Wahrheit nicht ernst nimmt, dem kann
man auch in grossen Dingen nicht vertrauen.  --Albert Einstein on honesty

Scheme-reports mailing list