Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site godot.UUCP
Path: utzoo!linus!decvax!cca!godot!bruce
From: bruce@godot.UUCP (Bruce Nemnich)
Newsgroups: net.lang
Subject: Re: FORTH efficiency
Message-ID: <146@godot.UUCP>
Date: Sat, 30-Jun-84 00:09:05 EDT
Article-I.D.: godot.146
Posted: Sat Jun 30 00:09:05 1984
Date-Received: Sun, 1-Jul-84 00:29:55 EDT
References: <289@harvard.ARPA>, <2224@mit-eddie.UUCP> <9623@lanl-a.UUCP>
Organization: Thinking Machines, Waltham, MA
Lines: 26

Forth is efficient in terms of space more than time.  I don't remember
the whole Forth history, but it goes back to the early 70s, when it was
used with early microcomputer chips to control industrial machines.  In
that day, 4k was a lot of memory.  Forth in incredibly compact, often
much more so than hand-optimized machine code.  I did an 8080
implementation once and remember writing an 8080 assembler for it which
took up about 600 bytes.

In terms of time-efficiency, it was much better than other interpreted
languages available and usually compared reasonably with the basic and
fortran compilers available for the larger 8-bit systems.  Yes, a good
compiler can usually create somewhat faster code (but not more
compact!).  Remember, the compilers for small machines were often
produced pretty primitive code, since the compiler itself had to be
small.  But most of all, forth is fun and convenient: having the
"compiler" and dislay editor and all the system commands resident is
great!  I also like it as a language a lot more than those two.

I have only used forth on small machines; its space advantages are pretty
meaningless in a large-address-space machine like the vax.  But I'll install
it, because its a language I like to play with.  It is also has a lot of
educational value.  Oops, shouldn't have said that!

-- 
--Bruce Nemnich, Thinking Machines Corporation, Waltham, MA
  {decvax!cca,ihnp4!mit-eddie,allegra!ias}!godot!bruce, BJN@MIT-MC.ARPA