🚀 go-pugleaf

[......]
Why are the contents of net.lang.forth cleared out so frequently?

Beats me.  Maybe it's not cleared out, but there's just little traffic.
Maybe we should try to spur some discussion.

In particular, like, whatsamatter with these standards commitees that
waste valuable time worrying about trivial details like what we use
for True/False flags or whether LEAVE bails directly out or sets
a flag for loop to end etc., instead of worrying about more useful
stuff like standardization for different word-size versions, Quan
variables, Native code versions, transient areas, developing a
minimum nucleus, and other stuff.  Like, the new features in Forth-79
and Forth-83 put together are so TRIVIAL, that I've not been inspired
to change from FIG-Forth, as it doesn't buy me enough to counteract
the trouble I'd have to go thru to change.

Guess we have to start our own standards commitee.  Here are some of
my propositions:


1. Standards for varying word-size versions.

   a. Defaults for ! @ , etc. are the word-size used.
   b. outside of that, C@,C!=8 bits, W@,W!=16 bits, D@,D!=32 bits
   c. The system constant WSIZE is defined as the number of BYTES
      in the system-default word.

2. Quan Variables.

   a. the word VAR is used to define these new variables
   b. a VAR definition is forced to be immeditate, as during compilation
      it will compile the correct code depending on the variable TOFLG.
      Given the sample variable named NAME,
      where TOFLG=              action performed is
	      0                  compiles equiv to NAME @
	      1                  compiles equiv to NAME !
	      2                  compiles equiv to NAME +!
	      3                  compiles equiv to NAME
	      4                  compiles equiv to NAME +
    c. The immediate words TO, +TO, ADDR, +ADDR are used to set
       TOFLG to 1-4 respectively.  After a compilation of a VAR word,
       TOFLG is set to the default of 0.


This is just a start.  I won't attempt to propose standards for native code,
transient areas, or a minimum nucleus at this time as these areas are complex
and for the moment, this message is long enough.

Just trying to spur some discussion.


Keith Doyle
{ucbvax,ihnp4,decvax}!trwrb!cadovax!keithd
"You'll PAY to know what you REALLY think!"

Keith Doyle suggested standards for Forth systems running on
machines with different word sizes.  Such a proposal may be found
in the Proceedings of the 1984 Rochester Forth Conference on
page 284.  The title is "Extended Addressing Wordset", by
Mitch Bradley and Bill Sebok.  A revised version of the proposal is
scheduled to be printed in the upcoming 32-bit machines issue of
the Journal of Forth Application and Research.

The proposal starts with a scheme similar to that mentioned by
Keith in his recent posting.  It also considers machines with
addresses larger than the stack width and stack manipulation operators.

Mitch Bradley