[antlr-interest] unable to run or debug grammar in antlrworks 1.4 on Windows XP

Matt Palmer mattpalms at gmail.com
Sat Jul 3 14:11:29 PDT 2010


Hi William,

well, in terms of grammar design, you're probably right that bitmask should
probably be a rule rather than a lexer token.  HEX_BYTE is a fragment, used
by other lexer tokens too.

However, the real issue for me is that the debugger sometimes works, and
sometimes doesn't,for the same grammar.  My only fix at this point is
commenting things out and back in until it starts mysteriously working!

cheers,

Matt

On 3 July 2010 22:05, William Clodius <wclodius at los-alamos.net> wrote:

>
> On Jul 3, 2010, at 2:58 PM, Matt Palmer wrote:
>
> > Well, this gets wierder, and to my mind seems to point to a bug in
> > Antlrworks itself.
> >
> > I verified that I could consistently run the debugger using one simple
> > grammar, but not another.  On the non-working grammar, I commented out
> rules
> > and tokens (starting from the most complex), until, lo-and-behold, after
> one
> > rule went the debugger started working.  I then re-enabled all the other
> > rules one by one, and the grammar kept working.  The lexer rule that
> wasn't
> > working was  a literal followed by a fragment, looking like this:
> >
> > BITMASK :   '&' HEX_BYTE ;
> >
> > I then re-enabled this rule... and the debugger kept working.  The only
> > difference at this point between where I started with a non-working
> debugger
> > and where I finished with a working debugger is I moved a couple of rules
> to
> > the end (comments and whitespace both going to hidden channels) to
> simplify
> > commenting them out.
> >
> > I've faced this problem before, I still have no idea what causes it and
> how
> > to fix it.  Maybe this will make sense to someone.
> >
> > Matt.
> <snip>
>
> Normally I would expect bitmask to be a syntactic element not a lexical
> element. If so then change all occurrences of BITMASK to bitmask. Otherwise
> have youm marked HEX_BYTE as a fragment?
>
> List: http://www.antlr.org/mailman/listinfo/antlr-interest
> Unsubscribe:
> http://www.antlr.org/mailman/options/antlr-interest/your-email-address
>


More information about the antlr-interest mailing list