[antlr-interest] Problems debugging large grammar in ANTLRWorks

[Jim Idle]

I meant to add, that I personally get more mileage from just using
Visual Studio 2005 to debug the actual generated code - tends to give
you more of the information that C programmers are used to, even if you
don't get the 350Mb graphic display ;-)

Jim

> -----Original Message-----
> From: antlr-interest-bounces at antlr.org [mailto:antlr-interest-
> bounces at antlr.org] On Behalf Of troy runkel
> Sent: Friday, June 01, 2007 11:32 AM
> To: ANTLR mail-list
> Subject: [antlr-interest] Problems debugging large grammar in
> ANTLRWorks
> 
> Greetings,
> 
> I'm working with a fairly large grammar (1620 lines) and was hoping to
> debug it using ANTLRWorks 1.0.1.  When I select the 'Debug...' option
> from the Debugger menu I get the following error message popup.  My
> grammar is using the language=C option.
> 
>      Error while generating the grammar:
>      (10): internal error: group CDbg line 33: template outputFile has
> no region called imports
> 
> This looks like a problem with the C code generator so I commented out
> the language=C option line and tried again.  After that I get the
> following error message in the Console window.
> 
>      [12:22:02] \tmp\antlrworks\TSQLParser.java:20: code too large
>      [12:22:02]     public static final String[] tokenNames = new
> String[] {
>      [12:22:02]                                  ^
>      [12:22:02] 1 error
> 
> It looks like maybe there are too many token strings in the tokenNames
> array?  Does anybody have any advice on what I need to do to get my
> grammar working with the ANTLRWorks debugger?
> 
> I also noticed that after processing the grammar ANTLRWorks is
> consuming over 900MB of memory on my Windows XP desktop machine.  If I
> avoid ANTLRWorks and process the grammar directly with ANTLR I get up
> to around 600MB of memory.  Is this normal?  Does that mean there's an
> upper bound to the size of the grammar that ANTLR can process?  I've
> been running java with the -Xmx1490m option to avoid running out of
> java heap space.
> 
> In addition to the language=C option I'm also using backtrack=true and
> memoize=true.
> 
> Thanks.
> 
> Troy Runkel