Path: utzoo!attcan!uunet!lll-winken!claris!hearn
From: hearn@claris.com (Bob Hearn)
Newsgroups: comp.lang.misc
Subject: Re: Language Design
Message-ID: <9091@claris.com>
Date: 17 Mar 89 20:50:24 GMT
References: <5200040@m.cs.uiuc.edu> <12443@watdragon.waterloo.edu>
Reply-To: hearn@claris.com (Bob Hearn)
Organization: Claris Corporation, Mountain View CA
Lines: 27

>> (For example, in the language I'm learning now, C++, if you forget to 
>> declare a class, it is a *syntax* error.  Not misuse of identifier or 
>> undefined class, *syntax* error.)
>> 
>
>Don't blame C++.  Blame C.  
>
>C is not context-free parsable [actually I believe it is, but not 
>with any sensible grammar].
>-- 
>Gordon V. Cormack     CS Dept, University of Waterloo, Canada N2L 3G1
>gvcormack@waterloo.EDU  gvcormack@uwaterloo.CA  gvcormac@water.BITNET

Well, the *grammar* defining C programs is certainly context-free!  Just
look in the back of K&R.  This doesn't mean that any grammatically correct
program is semantically correct.  I don't know of any non-trivial 
language with a context-free grammar for semantically correct programs.  (I
may be wrong here, but certainly this is correct for most languages.)
Actually, I've always thought parts of the grammar (e.g. type declarations)
were hacked to facilitate easy parsing.
Many compilers, however, do consider some semantic errors syntax errors,
because it's easier that way.  As for the C++ syntax error described above,
that's compiler-dependent.  MPW C++ tells me that the identifier is not
a valid type name.

Bob Hearn
hearn@claris.com