Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!decwrl!decvax!ima!compilers-sender
From: nigelh@uvicctr.UVic.ca.UUCP (R. Nigel Horspool)
Newsgroups: comp.compilers
Subject: Re: Dynamic Operators in Prolog
Message-ID: <653@uvicctr.UVic.ca.UUCP>
Date: 26 Apr 89 15:30:49 GMT
References: <3780@ima.ima.isc.com>
Sender: compilers-sender@ima.ima.isc.com
Reply-To: nigelh@uvicctr.UVic.ca.UUCP (R. Nigel Horspool)
Organization: University of Victoria, Victoria B.C. Canada
Lines: 29
Approved: compilers@ima.UUCP

I have managed to implement a parser for Prolog and its
dynamically-defined operators using lex & yacc, but it is
extremely messy.  Since all the shift-reduce conflicts are
resolved at parser-creation time, there are only two possible
approaches.

1.  Construct a grammar where all possible precedence levels for the
    operators are pre-defined.  (But typical Prologs allow 255 or more
     precedence levels.)

2.  Have semantic actions attached to production rules that
    look up the operator precedences and associativities.
    The semantic actions then select a shift action or a reduce
    action by telling the lexer to insert extra symbols into the input.
    The extra symbols subsequently cause the parse to follow either a
    shift path or a reduce path.

I implemented #2, but this simple idea has a lot of complications.
I will happily e-mail a short paper giving the full details to
anyone who requests it.


R. Nigel Horspool
nigelh@csr.uvic.ca, nigelh@uvunix.bitnet
--
Send compilers articles to compilers@ima.isc.com or, perhaps, Levine@YALE.EDU
Plausible paths are { decvax | harvard | yale | bbn}!ima
Please send responses to the originator of the message -- I cannot forward
mail accidentally sent back to compilers.  Meta-mail to ima!compilers-request