[EMAIL PROTECTED] wrote:
>
> ... First, here's a version of ROUND that can handle values
> where to-integer chokes, fractions less than one, and negative numbers:
>
> round: func [n [number!] p [integer!] /local factor neg] [
> if negative? n [n: (- n) neg: true]
> factor: 10.0 ** p
> n: n * factor + 0.5
> n: n - (n // 1) / factor
> either neg [- n][n]
> ]
>
Excellent! Thanks for replacing that top-of-the-head QAD with some
real code!
>
> Note that FORMAT has a refinement /full which won't work without the
> function FULL-FORM. I've been working on this with Larry Palmiter and
> Gerald Goertzel. It'll be available on rebol.org in a couple of days.
> FULL-FORM is meant to give the simplest possible string representation
> of a decimal value that will load back to exactly the same value.
>
> >> form pi
> == "3.14159265358979"
> >> pi - load form pi
> == 3.10862446895044E-15 ; not equal!
> >> full-form pi
> == "3.141592653589793"
> >> pi - load full-form pi
> == 0 ; equal!
>
This sounds seriously non-trivial. I'm looking forward to it.
-jn-
