[antlr-interest] Manual tree construction in 2.7.6 (buggy?)

[F Reig]

On 3/28/06, Micheal J <open.zone at virgin.net> wrote:
> > I construct the tree by hand like this:
> >
> > decl_default!
> >     : d:DEFAULT i:IDENT COMMA e:exp SEMI
> >
> >        {## = #(#d, #([IF,":IF"],
> >                      #([EQ_EQ,"=="],#i,#([NIL,"NIL"])),
> >                      #([STMTS,"STMTS"],
> >                            #([EXPR,"EXPR"],
> > #([BECOMES,"BECOMES"],#i,#e))
> >                       )
> >                     )
> >               );
> >        }
> >     ;
> >
> > When I dump the tree (or visualize it with antlr.astframe) I see this:
> >
> > ( :DEFAULT ( :IF ( == x NIL 1) ( STMTS ( EXPR ( BECOMES x NIL 1 )))))
> >
> > That is, there is a spurious "1" in the test, and a spurious
> > "NIL" in the assignment. Yet, the tree constructor looks OK
> > to me. Am I doing something wrong
>
> Yes. You are trying to add the single IDENT node to two different AST
> [sub-]trees.
>
> You need to copy the node if you wish to use it multiple times. Use
> dupTree() for that and be sure to set the first-child and next-sibling links
> appropriately.
>
> Cheers,
>
> Micheal

This seems to work:

         { AST i2 = astFactory.dupTree(#i);

         ## = #(#d, #([IF,":IF"],
                      #([EQ_EQ,"=="],#i,#([NIL,"NIL"])),
                      #([STMTS,"STMTS"],
                            #([EXPR,"EXPR"],
                                #([BECOMES,"BECOMES"], i2 ,#e)))));

And it seems that I don't need to set any links by hand. Thanks Micheal

I would suggest that a similar example be included in the manual.

Fermin