(Message perl/py:5)
Replied: Wed, 22 Apr 1998 10:40:39 -0400
Return-Path: gbacon@mickey.cs.uah.edu
Return-Path: <gbacon@mickey.cs.uah.edu>
Delivered-To: mjd-filter-deliver@plover.com
Received: (qmail 30929 invoked by uid 119); 22 Apr 1998 14:39:13 -0000
Delivered-To: mjd-filter@plover.com
Received: (qmail 30924 invoked by uid 119); 22 Apr 1998 14:39:09 -0000
Delivered-To: mjd@plover.com
Received: (qmail 30921 invoked from network); 22 Apr 1998 14:39:07 -0000
Received: from pluto.cs.uah.edu (146.229.2.78)
  by plover.com with SMTP; 22 Apr 1998 14:39:07 -0000
Received: from mickey.cs.uah.edu (from mickey [146.229.2.62]) by pluto.cs.uah.edu (SMI-8.6/SMI-SVR4)
	id JAA01846; Wed, 22 Apr 1998 09:30:19 -0500
Message-Id: <199804221430.JAA01846@pluto.cs.uah.edu>
To: Mark-Jason Dominus <mjd@plover.com>
Subject: Re: possible py bug 
Reply-To: gbacon@cs.uah.edu
X-Face: "7sfK9d*!trQg^>swxP}M|R|<7$tFtlm|Z(4M$i`?}l87xigz\37|mQUBi/ct#>=H,&(V!A
        6=jnjOfA#.0Q@jF1t~Eue)5=zFI
In-Reply-To: Your message of "Wed, 22 Apr 1998 10:21:53 EDT."
             <19980422142153.30847.qmail@plover.com> 
Date: Wed, 22 Apr 1998 09:29:40 -0500
From: Greg Bacon <gbacon@mickey.cs.uah.edu>

In message <19980422142153.30847.qmail@plover.com>,
    Mark-Jason Dominus writes:
: Actually, could you just send me the original .y file, and
: the test input that you gave to the parser?

---- begin assem.y ----
%token ID INT

/* directives */
%token START

/* instructions */
%token ADD

%%

line    : label direct
        | label instr
        ;

label   : ID
        |
        ;

direct  : START addr
        ;

instr   : ADD addr
        ;

addr    : INT
        | ID
        ;
---- end assem.y ----

---- begin source.asm ----
TEST    START   0
        ADD     TEST
---- end source.asm ----

In case you're wondering, I'm writing an assembler for a system software
class.

: > I changed line 60 of py-skel.pl to
: > 
: >     $state = $states[$#states] || 0;
: > 
: > and it worked, 
: 
: I bet it didn't.  For one thing, I'm totally sure line 50 is wrong:
: -      @args = splice(@values, -$length[$d]);
: +      @args = splice(@values, -$length[$d]) if $length[$d]
: 
: Otherwise, you splice out the entire value stack.

TOKENS: [][TEST][       ][START][       ][0]
$VAR1 = {
          'line' => 1,
          'label' => 'TEST',
          'direct' => {
                        'addr' => '0',
                        'direct' => 'START'
                      }
        };

TOKENS: [       ][ADD][ ][TEST]
$VAR1 = {
          'line' => 2,
          'instr' => {
                       'instr' => 'ADD',
                       'addr' => 'TEST'
                     }
        };

...which is the correct parse.

: > Is there any equivalent of yyerror (or how do I say `fail' from within
: > a rule)?
: 
: Do you mean YYERROR?  (I haven't done this in a really long time.)  I
: don't think so but it shouldn't be hard to add.  Let me reread the
: doscs and get back to you.

What I'm interested in doing is purposefully parsing bad syntax to give
error messages that are a little more helpful.  I suppose it would be
possible to examine the result from yyparse's caller to achieve the same
effect.  It would still be nice to have a way to abort the parse from
within the rules.

Greg