Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 beta 3/9/83; site nbs-amrf.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!harpo!seismo!rlgvax!cvl!umcp-cs!nbs-amrf!libes
From: libes@nbs-amrf.UUCP
Newsgroups: net.lang.apl
Subject: Re: Sharing APL programs
Message-ID: <284@nbs-amrf.UUCP>
Date: Sat, 19-May-84 16:12:09 EDT
Article-I.D.: nbs-amrf.284
Posted: Sat May 19 16:12:09 1984
Date-Received: Mon, 21-May-84 03:41:55 EDT
References: <198@hou2f.UUCP>, <7644@watmath.UUCP>, <7681@watmath.UUCP>
Organization: National Bureau of Standards
Lines: 31

I was never too involved in commercial APL usage so I don't know if
the WorkSpace Interchange Standard was ever really used.  I wrote one
for VSAPL a long time ago (ask CCIS at Rutgers University) and I know
DEC had one for APLSF.  I don't know anyone besides myself that ever
used the WSIS.  It was good (as far as I was concerned) for storing
backups to tape.  I suspect that STSC (and probably all the major
APL vendors) have it since I used their technical report when I wrote
it.

It certainly isn't applicable for transferring functions via netnews.
The way it worked was to produce a supposedly independent
representation of characters and then generate indices of the
characters (at the source code level).  You would then read it in
these character independent indices and generate a site dependent APL
function.  In some ways it was very complete (it could handle 17-bit
characters if thats what you had) and hence very verbose in terms of
storage requirements.  In other ways it was somewhat useless - it did
not address issues like standard file access methods and hence you
almost always had to get your hands dirty when moving programs from
one vendors system to another.

I consulted for a company (many years ago) that moved all their APL
code from one APL system to another.  Getting the character codes
correct was no problem.  The time consuming part was recoding when we
found out subtle differences between many of the functions.  Also, the
former system could execute system commands and the latter couldn't.
And you could code in assember in the former whereas you couldn't in
the latter.  I learned quick that APL is not machine independent in
practice.

Don Libes		{seismo,allegra}!umcp-cs!nbs-amrf!libes