Icon

[utzoo!duke!unc!whm]

It has been stated that Icon from the U of Ariz. is the research successor
to Snobol, but as I understand it is also to be the production sucessor
to Snobol.  Icon seems to be between the high-powered capabilities
of Snobol 4 with dynamic everything and the strongly typed "modern" languages
like Pascal and Ada.  Icon has the full power of Snobol with respect to
pattern matching (all of the Snobol pattern primitives can be constructed
in a couple of pages of Icon), while having the control structures commonly
found in most languages and not in Snobol.  With the advent of Icon 3, which
runs on 11's with i/d, Icon seems to be a language well suited to the
Unix environment.  And now, with Paul Eggert's (Santa Barbara) adaption of
Icon 3V for Berkeley Unix, all of the Unix Vaxen can have this great system.

Typical compile and link times for Icon on a Vax run around 10-20 sec for
small (50-100) line programs.  However, most of these programs may replace
a C program of several hundred lines, so for programming when you are doing
a lot of string handling, Icon seems like the way to go.

I have done a distribution package for IBM 370's and Vax/VMS systems for
the portable Icon 2 systems, and have found Icon to be a fun language to
work with.  We also have the Unix Icon package for our Vax (which unfortunately
only runs Unix on rare occasions) and I am planning to install the same
package under Eunice under (yuck) VMS.

					Bill Mitchell
					N.C. State University
p.s. Does anybody in the world except me use RATSNO? (Rational Snobol).