Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!unix.cis.pitt.edu!pitt!willett!ForthNet
From: ForthNet@willett.pgh.pa.us (ForthNet articles from GEnie)
Newsgroups: comp.lang.forth
Subject: ANS TC Magnet for LOOPS, EXIT, Term.
Message-ID: <2126.UUL1.3#5129@willett.pgh.pa.us>
Date: 17 Dec 90 00:23:37 GMT
Organization: String, Scotch tape, and Paperclips.  (in Pgh, PA)
Lines: 52

Category 10,  Topic 16
Message 94        Sat Dec 15, 1990
F.SERGEANT [Frank]           at 11:08 CST
 
     Yesterday I sent in the following proposal.  If I have made any  errors
in the facts or history of FOR NEXT please let me know:
================================================================= ANSI X3J14
FORTH TECHNICAL PROPOSAL     ----------------------------------- TITLE:  FOR
... NEXT Should Execute the Body Count Not Count+1 Times  PROPOSAL: 
     If  FOR ... NEXT  and its index or count are mentioned in the  Standard,
make the loop body execute u times rather than u+1 times.
     For example:
          : TEST1  ( u -)  FOR    ." A"    NEXT    ;
          0 TEST1  should print nothing at all
          3 TEST1  should print  AAA

          : TEST2  ( u -)  FOR    I .      NEXT    ;
          0 TEST2  should print nothing at all
          3 TEST2  should print   2 1 0     DISCUSSION:
     As short-hand for the sake of discussion, the version I'm  proposing will
be called "superior" and the u+1 version will be called  "inferior."
     The (short and thin) tradition of having the body of a FOR NEXT  loop
execute u+1 times for a non-zero count and execute one time for a  zero count
stems strictly from hardware efficiency considerations on  the NOVIX and
originated with cmFORTH for the NOVIX.  From there it was  carried over to the
RTX.
     Rob Chapman in his botFORTH (on RTX & 68000 machines) and I in my  Pygmy
Forth (on 80x86 MS-DOS machines) have already converted to the  superior FOR
NEXT.  Our experience with this is favorable.  I  understand from Rob that he
discussed this possible change with Charles  Moore at the 1990 FORML and that
CM had no objection to the superior  version.
     When the body always executes at least once and executes u+1  times,
programs very often must use an awkward work-around to test for  the zero
count in order to bypass the loop.  The following phrase is common, and
wasteful, and ugly:
           ?DUP IF   1- FOR ... NEXT   THEN with the new version the above
phrase would be
           FOR ... NEXT Clearly superior!  In cases, where you want the body
to execute u+1  times, that is easily achieved with the superior version by
saying
          1+ FOR ... NEXT
     The "tradition" of the inferior version is far too short to  justify
embedding it into the Standard.
     In those special cases where a little extra performance can be  wrung out
of a Forth "engine" by using the inferior version, vendors  are still free to
define it, but with a different name. DATE: 14 December 1990.  SUBMITTED BY:
                             Frank Sergeant
                             809 W. San Antonio St.
                             San Marcos, Texas  78666
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: dwp@willett.pgh.pa.us or uunet!willett!dwp