Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!mit-eddie!uw-beaver!teknowledge-vaxc!sri-unix!garth!smryan
From: smryan@garth.UUCP (Steven Ryan)
Newsgroups: comp.lang.misc
Subject: Re: Ada and C and Algol 68
Summary: Those who cannot learn from the past must...those who cannot learn from
Message-ID: <771@garth.UUCP>
Date: 21 Jun 88 20:04:26 GMT
References: <7997@alice.UUCP>
Reply-To: smryan@garth.UUCP (Steven Ryan)
Organization: INTERGRAPH (APD) -- Palo Alto, CA
Lines: 23

I detect echoes of +:= *:= et cetera.

A central principle in programming language design is don't look at previous
work and avoid the landmines.

A case in C was =+ =* =- and so forth. If the Revised A68 report chapter 9
had been studied with care, the ambiguity would have been immediately noticed.

A case in Ada had to do with what every they call structure display. The long
form is type-name(member1,...,membern) and short form is (member1,...,membern).
If the short form only has one member, (member), it is ambiguous with a
parenthised expression. Again the A68 directly dealt with this problem by
requiring a collateral clause to have zero or more than one unit.

One of the nicest things about the Revised Report was a complete and formal
definition. It might take a bit work, but every question could be resolved.

Why bother? Isn't much more fun to let each compiler have its own
interpretation? comp.lang.c and comp.lang.ada consume a large bandwidth as
people try to decide what is and is not legal and what it means.

(I take it back--I have been informed C does have a formal definition:
"Whatever PCC does on a VAX.")