[antlr-interest] AST loosing leaves in 3.3 vs. 3.2

[Joseph Cottam]

Thanks for the help.  I didn't know the behavior I was employing was
undefined.  With some re-factoring I have my grammars working again.  The
refactor was more complex when the sub rule could return different root
types (I ended up adding sub-rules with predicates) and when the contents of
the new root-node depend on a value matched in the sub-rule (where I now use
a return value).  It works great though, I'm glad to know now that I was
mistaken before.

-JC


On Tue, Jan 4, 2011 at 6:54 PM, Jim Idle <jimi at temporal-wave.com> wrote:

> Your grammar is erroneous and just 'happened' to work in 3.2. The root of
> a node can only be the result of a sib rule if the subrule contains a
> single node, otherwise the results are undefined.
>
> You would need:
>
> callTarget
>  : call
>     -> ^(FUNCTION call PACK)
>  ;
>
> call
>   : ID ID ID
>   ;
>
> Which is easily constructed in all situations, even if you add an extra
> root node:
>
>
> callTarget
>  : call
>     -> ^(CALL call PACK)
>   ;
>
> call
>  : one=ID two=ID three=ID
>    -> ^(FUNCTION $one $two $three)
>  ;
>
>
> > -----Original Message-----
> > From: antlr-interest-bounces at antlr.org [mailto:antlr-interest-
> > bounces at antlr.org] On Behalf Of Joseph Cottam
> > Sent: Tuesday, January 04, 2011 3:48 PM
> > To: antlr-interest at antlr.org
> > Subject: [antlr-interest] AST loosing leaves in 3.3 vs. 3.2
> >
> > Upgrading from v3.2 to v3.3, I started loosing the leaves in some of my
> > rules.  All affected rule-pairs have the same structure: The results of
> > called rule becomes the root of the AST returned by the rule doing the
> > calling AND the calling rule adds more nodes to the AST (not just
> > returning the results of the called rule).  In the example below, the
> > result of matching the "call" rule is truncated to just its root when
> > it is referred to in the construction of the AST for "callTarget."  I
> > feel like I am missing something simple, but I can't figure out what
> > part of the v3.3 release notes covers this circumstance.
> >
> > 3.2 output: (FUNCTION first second third PACK)
> > 3.3 output: (FUNCTION PACK)
> >
> > -Joseph A. Cottam
> >
> >
> > input------------------------
> > first second third
> >
> > grammar-------------------
> > grammar Error;
> >
> > options {
> >   language = Java;
> >   output=AST;
> > }
> >
> > tokens {FUNCTION; PACK;}
> >
> > callTarget
> >  : call
> >     -> ^(call PACK)
> >  ;
> >
> > call
> >   : one=ID two=ID three=ID
> >     -> ^(FUNCTION $one $two $three)
> >   ;
> >
> > ID    : ('a'..'z' | 'A'..'Z' | '_') ('.'? ('a'..'z' | 'A'..'Z' | '_' |
> > '0'..'9'))*;
> > WS  : (' '|'\r'|'\t'|'\u000C'|'\n')+ {skip();};
> >
> > List: http://www.antlr.org/mailman/listinfo/antlr-interest
> > Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-
> > email-address
>