Newsgroups: comp.robotics
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!fas-news.harvard.edu!newspump.wustl.edu!news.starnet.net!wupost!howland.reston.ans.net!ix.netcom.com!netcom.com!jfox
From: jfox@netcom.com (Jeff Fox)
Subject: Re: Forth's reputation
Message-ID: <jfoxD43pDp.6yq@netcom.com>
Sender: jfox@netcom7.netcom.com
Organization: Netcom Online Communications Services (408-241-9760 login: guest)
References: <6.14254.723@spacebbs.com> 
             <jfoxD42n06.7CB@netcom.com> <nagleD42tCK.7vp@netcom.com>
Date: Thu, 16 Feb 1995 16:20:13 GMT
Lines: 58

In article <nagleD42tCK.7vp@netcom.com> nagle@netcom.com (John Nagle) writes:

>     Forth chips have an amusing history.  The first one by Moore
>had an error in the multiply unit (the product always came out even)
>and he couldn't afford to have a new rev made.  The literature thus
>pointed out that, after all, most multiplies have at least one even
>operand.  (No, I'm not making this up.)

It also had a bug in the interupt logic that was never fixed, although
a pal was developed to get interupts working, and of course there was
a software fix for the multiply bug in Novix.  

>     He also offered a development system that used a 3-button keyboard
>as input.  That's all you had; three buttons on which you could do
>chords.  It was actually possible, although painful, to program with this.

No one but Chuck ever used a 3-button chord keyboard to my knowledge.  The
kit software used a serial cable to a PC for software development.  But I
have always thought it strange the Chuck seemed to get more publicity for
his keyboard than he did for making a chip that was an order of magnitude
faster than anything from Intel at the time from only 4000 gates.

>     Harris eventually took over the product and made one that worked
>right, selling it into the embedded systems market.  It may still be sold.
>But it was never a big seller.

It was more popular in Europe than in the US.  This may have had something
to do with the perception in the US that RAM is cheap and efficiency is
not a factor.  It is funny, with RAM is short supply, and being a big
source of money flowing out of the US, people have not yet realized that
DRAM has become the gasoline of the 90s.  Our cars are getting better
gas economy, but our computers now need 16megabytes of DRAM to be able
to write a letter according to the "shopping channel."  People seem to
think this proping up our computer industry, when it is really a
big trade balance problem.

>     Forth is a nice language for embedded programs that are close to
>the hardware.  I've used the New Micros Forth for the 68HC11, which is
>OK, although on the slow side, for a mobile robot with a steerable
>sonar.  But maintaining large Forth programs written by others is 
>considered very painful.  Despite that, sizable applications (including
>Swivel-3D for the Macintosh) have been written in Forth.
>
>					John Nagle

Maintaining even small Forth programs can be painful if they are not
written with that in mind.  The freedom do most things any way you want
combined with a lack of education has often led to true stories about 
Forth being "write only".  However there are many large applications,
written by large teams of Forth programmers that have been very sucessful
and proven quite maintainable.  However these projects often involved
getting professional Forth training for the programming team, and also
proper management of the project.  Forth Inc., has trained teams of
people at IBM, JPL, NASA, etc for some of those large projects.  Yet
few people are aware of the things that are offered or are possible
with the kinds of systems that they offer.

Jeff Fox
