So, this works to set RD_TRACE:
        perl -MParse::RecDescent -s - -RD_TRACE grammar NewMakepp::Grammar

Unfortunately, it looks like the path taken by the precompilation fails to 
actually check for undefined rules.  This may be a bug, or may have been 
intentional, I'll have to follow up with Damian.

Even if this were fixed, I'm not sure how useful the '-s' command line option 
is, as -RD_AUTOSTUB will set $::RD_AUTOSTUB to '1'.  An undocumented feature of 
$::RD_AUTOSTUB is that if it has a boolean-true value, an undefined rule named 
'foo' will not get the production
        foo: 'foo'
but instead get the production:
        foo: $::RD_AUTOSTUB
Which, for -s -RD_AUTOSTUB, will generate:
        foo: '1'
        1: '1'
Which probably isn't what you wanted in the first place.

Long story short, RD_AUTOSTUB doesn't currently work with precompilation of 
parsers.  You'll have to either not precompile, or define your own stub 
productions.

Please file a ticket for this issue.  I'm out of town for the next week, but 
will try to get this into a bugfix release next week.  I've included details on 
the issue below, feel free to include them in the bug.  You can find the ticket 
tracker for Parse::RecDescent here:
https://rt.cpan.org/Public/Dist/Display.html?Name=Parse-RecDescent


Note that this isn't unique to the command line, but appears to be a problem with the Precompile method. The callstack is something like:
        script:     Parse::RecDescent->Precompile($grammar, $class, $sourcefile)
        Precompile: $self = Parse::RecDescent->new($grammar,1,$class)
        new:        $self->Replace($grammar, 1, $class)
        Replace:    _generate($self, $grammar, 1(fromsplice), 1, $class);
        _generate:  my ($self, $grammar, $replace, $isimplicit, $isleftop) = 
(@_, 0);

So, the '1' added to @_ from Precompile in the call to new() ends up becoming 
$isimplicit.

$isimplicit prevents $self->_check_grammar() from being called, which is the 
function that does the RD_AUTOSTUB checking/generation.  I'll have to look into 
why _check_grammar is skipped for $isimplicit cases.

There also appears to be another bug here, where the $class parameter gets 
incorrectly propagated from Precompile into $isleftop inside of _generate, 
which is probably not correct.  Unless I've misread the code somewhere.


On 1/12/2012 10:01 AM, Yuri Shtil wrote:
Well, I found as much myself, tried all I could and it did not work.

-----Original Message-----
From: jtbr...@cpan.org [mailto:jtbr...@cpan.org]
Sent: Thursday, January 12, 2012 9:53 AM
To: recdescent@perl.org
Subject: Re: How to build parser with autostubbing

Consult the section on '-s' at http://perldoc.perl.org/perlrun.html:

     -s

     enables rudimentary switch parsing for switches on the command line
     after the program name but before any filename arguments (or before
     an argument of --). Any switch found there is removed from @ARGV and
     sets the corresponding variable in the Perl program. The following
     program prints "1" if the program is invoked with a -xyz switch, and
     "abc" if it is invoked with -xyz=abc.
          #!/usr/bin/perl -s
          if ($xyz) { print "$xyz\n" }

     Do note that a switch like --help creates the variable ${-help} ,
     which is not compliant with use strict "refs" . Also, when using
     this option on a script with warnings enabled you may get a lot of
     spurious "used only once" warnings.


So this is probably what you want:

        perl -MParse::RecDescent -s -RD_AUTOSTUB - grammar NewMakepp::Grammar

Untested, you may need to move -s, '-RD_AUTOSTUB  and the '- grammar ...' 
arguments around to make it work.

On 1/12/2012 9:39 AM, Yuri Shtil wrote:
The documentation is a little unclear

I use perl -MParse::RecDescent - grammar NewMakepp::Grammar to create a parser 
module.

What do I do in order to generate stubs for undefined rules?
How do I pass RD_AUTOSTUB?

----------------------------------------------------------------------
------------- This email message is for the sole use of the intended
recipient(s) and may contain confidential information.  Any
unauthorized review, use, disclosure or distribution is prohibited.
If you are not the intended recipient, please contact the sender by
reply email and destroy all copies of the original message.
----------------------------------------------------------------------
-------------


Reply via email to