Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!uunet!amgraf!cpsolv!rhg
From: rhg@cpsolv.UUCP (Richard H. Gumpertz)
Newsgroups: comp.lang.misc
Subject: Re: PL/I Grammar
Summary: PL/I is not LALR
Keywords: PL/I, C, grammar, parser
Message-ID: <422@cpsolv.UUCP>
Date: 21 Oct 89 02:06:16 GMT
References: <2956@usceast.UUCP>
Reply-To: rhg@cpsolv.uucp (Richard H. Gumpertz)
Organization: Computer Problem Solving, Leawood, Kansas
Lines: 28

In article <2956@usceast.UUCP> sharp@usceast.UUCP (Harry Sharp) writes:
>...  I have been told that PL/I cannot be parsed LALR.  
>Is this true? (I hope not.)

That is indeed true.  Many constructs cannot be recognized until after you
read past arbitrarily nested parentheses, thus killing any limited-lookahead
parser.

Another amusing problem is that there are NO reserved keywords.  One can even
declare variables with the same names as the builtin keywords!  Hence, with the
appropriate declarations preceding, the PL/I loop
    DO DO = TO TO BY BY WHILE (BEGIN = END) WHILE (BEGIN = END);
      BEGIN; ... END;
    END;
is legal.  For those not familiar with PL/I, it is roughly equivalent to the
C loop
    for (increment = WHILE(BEGIN == END), limit = BY, *(index = &DO) = TO;
         (increment >= 0 ? *index <= limit : *index >= limit) & (BEGIN == END);
         *index += increment)
      { ... }
where index, increment, and limit are automatically declared with appropriate
type.  The order of evaluation of the four assignments on the first line is not
defined by the PL/I language.  No, the & in the second line shouldn't be an &&.
-- 
==========================================================================
| Richard H. Gumpertz   rhg@cpsolv.UUCP -or- ...!uunet!amgraf!cpsolv!rhg |
| Computer Problem Solving, 8905 Mohawk Lane, Leawood, Kansas 66206-1749 |
==========================================================================