Re: "The Best Programming Language for the End of the World"

On 9/04/2025 6:42 pm, Hans Bezemer wrote:

On 09-04-2025 04:21, dxf wrote:

On 9/04/2025 1:15 am, dxf wrote:

...
It's not SET-PRECISION (presumably the implementer has made that sane)
but rather variability when values are sane e.g.
>

6 set-precision  ok
1e-4 f.  0.000100000  ok
>
6 set-precision  ok
1e-4 f. 0.0001  ok

>
While both of those are numerically correct, they're plainly different i.e.
11 characters vs. 6 characters.  As an app writer I'd like some uniformity
here as I might want to output several columns of numbers on a screen.  To
do that I need to know how wide to make those columns.
...

>
Iforth's F. was previously mentioned.  In addition to treating PRECISION as
a decimal points specifier (ok it was their decision) F. includes mode
switching based on something called 'fieldwidth' for which I could find no
reference.  Is mode switching good or bad?  As always - it depends.  I've
yet to see a bank statement with the balance displayed in scientific
notation.  (But then I'm not Elon Musk so who knows.)
>
I mention this to demonstrate the variability that exists across forth -
presumably because Forth-94 chose not to think it through leaving folks
to their own devices.  REPRESENT ?  Well, I could rant about that too :)

 
I've never been an avid FP wordset user, so I must admit I've never delved so deep into it. But when I did (last night) I thought "Darn, he's right!"
 
1. I remember my spreadsheet having a way to set the number of decimals. So I looked there. It used F.R to use that result - and to my surprise (it seems .R little brother) AFAIK it's not part of the FP wordset;
 
2. So I looked in the standard (obvious when you're coming from F.R) and no - nothing special there: "Set the number of significant digits". Well that's helpful. But do you mean "significant digits" the way I mean "significant digits"? I don't think so.. The beauty of "most significant digits of the significand" didn't escape me, though. Merriam Webster, "significant": "of a noticeably or measurably large amount". Almost like  "most", you mean. ;-)
 
3. If I can't find any help there, try elsewhere. Like C++. There "Precision" really means "Precision" - aka "the number of significant digits in the mantissa". Which in my book means 19 for a 64-bit mantissa and 9 for a 32-bit mantissa - if someone is unsure of what I mean. "fixed" however, is the keyword in C++ that switches this meaning to the number of decimals. If I understand you correctly, that's how iForth actually works.
 
I see two ways out of this:
a. Add F.R to the standard;
b. Add SET-DECIMALS or SET-FIXED to the standard.
 
And finally, add a section to the Annex to define and clarify the whole darn thing (because IMHO that is dearly missing).

Also specify which of these should be displayed:

6 set-precision
0.000100000
0.0001       ( redundant trailing zeros removed)

For my money I'd have:

(F.) ( r n -- adr len )

From that an app or library can define F.R :

\ Print a string right-aligned in a field width characters
: S.R ( adr len width -- )  over - spaces type ;

: F.R ( r n width -- )  >r (f.) r> s.r ;

Or where the display mustn't exceed field-width ...

\ Per S.R but print asterisks if string exceeds width.  Assumes
\ adr len can be written to.
: ?S.R ( adr len width -- )
  2dup > if  nip  2dup [char] * fill  dup  then  s.r ;

: ?F.R ( r n width -- )  >r (f.) r> ?s.r ;