Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site mh3bc1.UUCP
Path: utzoo!linus!philabs!seismo!harpo!eagle!mh3bs!mh3bc1!cw
From: cw@mh3bc1.UUCP
Newsgroups: net.sources
Subject: Re: Ada grammar
Message-ID: <110@mh3bc1.UUCP>
Date: Wed, 14-Dec-83 20:58:04 EST
Article-I.D.: mh3bc1.110
Posted: Wed Dec 14 20:58:04 1983
Date-Received: Fri, 16-Dec-83 03:16:26 EST
References: <671@ssc-vax.UUCP>
Organization: Bell Labs, Murray Hill
Lines: 25


David Norris has recently supplied a YACC version of an Ada grammar
to net.sources.  He claimed that the grammar is LALR(1).  It should
be noted that the grammar is not even LR(1).  He did note the
correct place where the problem occurs.  YACC will handle the problem
area by chosing to shift; Norris claims that this is a correct
resolution.  I have not tested the claim.

I also have not tested to see if the grammar in fact covers the
real Ada grammar.  This is a general problem with fast grammars for
Ada; to get them to be LALR(1) or LR(1), you actually have to
write a grammar for a little larger language.  Sometimes
you make a mistake and cut off a bit of Ada.  I have also not
checked to see if this grammar is appropriate for all of ANSI
1983 Ada.

We can, however, thank Norris for providing one more example
of a grammar that we can use to understand nice solutions to
Ada parsing problems.

Charles Wetherell
-- 
Charles Wetherell
MH 3C-327 x-3099
eagle!cw