Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!iuvax!ux1.cso.uiuc.edu!ux1.cso.uiuc.edu!uicsrd.csrd.uiuc.edu!jaxon
From: jaxon@uicsrd.csrd.uiuc.edu
Newsgroups: comp.lang.apl
Subject: Re: APL for SunWorkstations ?
Message-ID: <49700019@uicsrd.csrd.uiuc.edu>
Date: 16 Mar 90 17:01:00 GMT
References: <528@fwi.uva.nl>
Lines: 18
Nf-ID: #R:fwi.uva.nl:528:uicsrd.csrd.uiuc.edu:49700019:000:840
Nf-From: uicsrd.csrd.uiuc.edu!jaxon    Mar 16 11:01:00 1990


I love the idea that APL fits into a yacc grammar.

I take it that your APL dialect is not APL2-like nor ISO conforming!

We constructed our test suite (at Unisys) by collecting every example
in our manuals and quite a few from other vendors manuals.  In keeping
with this I published a few real monster case examples in the Unisys APLB
manual.  In particular the page describing the interaction between Execute,
the Each operator, the )SIS and error displays, the result-returning
potential of the executed string, and whether the result of the execute is
required or not.  

 (#IO is 1) rotate [ #IO is 0 ] 2 2 rho 'ABCD'

also raised some gray hairs.  I claim the result is 2 2 rho 'CDAB', the
axis qualifier being very tightly bound (before operands or arguments)
and being fully evaluated while still in an environment where #IO = 0.