Diagramming Robust AI

Dave Kenny dk at parka.winternet.com
Mon Mar 30 13:46:40 EST 1998


In comp.lang.forth Bart Lateur <bart.mediamind at tornado.be> wrote:
<snip>
: Of course, since then, an "official" 4th generation of languages was
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^
: created, which is, err.... either database query (like SQL), Prolog like
: languages, or maybe even something else (natural query language?).

To me the everyday meaning of "Fourth Generation Language" was stuff
like dBase or SQL;  i.e. special purpose and even application-specific
languages.  Except for SQL, these seemed to me to be hokey, proprietary
things more like a collection of library routines tied together by
yet another "basic-like" language.  Naughty, naughty trade-rags and
marketeers!

But there were a lot of good, innovative and powerful languages too,
that never got distinguished as "generations":

Besides machine code, assembly languages and the COBOL-FORTRAN-ALGOL
crowd, there were and are other languages that are both general and
vastly more expressive.  Lisp comes to mind (late '50's?).  Also APL
showed up in the 60's, and SNOBOL as well.  All very different and
very powerful languages.

What's weird about Forth is the relatively seemless connection it
provides from the human programmer to the hardware itself.  Another
oddity is that it abstracts away the notion of "paramter passing"
because it's not actually necessary:  much of the time it's just
unnecessary clutter.

So while Forth was a huge leap forward when it was created, it was also
a much more primitive language -- primitive in the sense of being
closer to the machine.  Forth achieves its power from its profound
simplicity, like unix did later, except that Forth takes this a
couple orders of magnitude further down.

I sometimes wonder what computing would be like had Forth been
discovered/invented much earlier.  Traditional assembly language
could be dispensed with.  Batch environments would be for running
payroll, NOT for doing development work.  Development work would be
fast and interactive, even on quite primitive hardware.





More information about the Neur-sci mailing list