"Bruce McFarling" <agila61@[EMAIL PROTECTED]
> wrote in message
news:5519e608-0eb9-44d8-abc3-f0cf1b40d096@[EMAIL PROTECTED]
> On Mar 30, 1:02 am, "Ed" <nos...@[EMAIL PROTECTED]
> wrote:
> > Note: this is an "old" version, which I only recommend for systems
> > which have the standard REPRESENT and can't otherwise upgrade.
>
> By upgrade, may I take it you mean has the features of REPRESENT as
> proposed in Represent_11.txt?
>
> ftp://ftp.taygeta.com/pub/Forth/Applications/ANS/Represent_11.txt
>
> That is, 0 is 0.00E+1, nearest rounding is used for representation,
> and a u of 0 gives the nearest rounding if the fp number handed to
> REPRESENT is the residual following the last significant place.
>
> Has that been put up for Forth200x? I'd vote for all of those, as well
> as a requirement that the buffer handed to REPRESENT be at least four
> characters.
That was a very early version. The current (and likely final) spec is:
ftp://ftp.taygeta.com/pub/Forth/Applications/ANS/Represent_21.txt
It requires apps allocate a minimum buffer size of MAX-FLOAT-DIGITS.
In practice most apps and forths already meet that. Apps that dynamically
ALLOCATE the buffer should ask for ( n ) MAX-FLOAT-DIGITS MAX chars.
I believe the spec is now solid and am leaving it for others to use as
they
see fit. I've briefly tried it out on Win32Forth and some commercial
systems.
--
Once the new REPRESENT is installed, systems may wish to replace
their existing f/p output functions with a full set similar to C, Fortran.
FPOUT (included in the spec) is public domain which means it can be
embedded, or its routines used, in a forth without licensing or
restriction of any sort.
|
