Fixed-point math, ANS, DPL and Win32Forth

Given that fixed-point math is 3 lines of code in Forth
((
10000 constant scale
: x* scale */ ;
: x/ scale swap */ ;
))
it is quite reasonable to have some built-in support for fixed point
number input.

Historically, there has been the variable DPL acquiring its value as a
side efect of number conversion.

For example,

1000 \ single, dpl=-1
1000. \ double, dpl=0
10.00 \ double, dpl=2

So,
1) Why DPL is not in the standard?
2) Why DPL is not in Win32Forth?
3) Is there an elegant way to implement fixed-point input in ANS?

I wrote fixed-point number input in ANS Forth, but it takes more lines
of code than the whole non-ANS (DPL-based) fixed point number support.


\ fixed-point number support
\ xn denotes a signed fiXed-point Number

: 10^ ( u -- 10^u ) 1 swap 0 ?do 10 * loop ;
4 constant logscale \ number of digits after decimal point

logscale 10^ constant scale
: x* ( xn1 xn2 -- xn3 ) scale */ ;
: x/ ( xn1 xn2 -- xn3 ) scale swap */ ;

: x ( d -- xn ) drop  logscale dpl @[EMAIL PROTECTED]
 0 max - dup abs 10^ swap 0< if /
else * then ;
: x. ( xn -- ) s>d tuck dabs <# logscale 0 ?do # loop [char] . hold #s
rot sign #> type space ;
: [x] ( "xn" -- ) ( run: -- xn ) postpone [ parse-word evaluate x ]
postpone literal ; immediate


\ the ANS version
: xx ( "xn" -- xn )
    parse-word over c@[EMAIL PROTECTED]
 [char] - = dup >r if 1 /string then
    0 0 2swap >number
    dup if
        over c@[EMAIL PROTECTED]
 [char] . <> abort" expected . inside a number" 1 /
string
        logscale min dup >r >number nip 0<> abort" canot convert" r>
    else
        nip
    then
    nip logscale swap - 10^ * r> if negate then
;
: [xx] ( "xn" -- ) ( run: -- xn ) xx postpone literal ; immediate



--
I post, therefore I am