[antlr-interest] TreeParser: Qs about generated code

[rustydstone]

Simple grammar:

expr returns [Object o] {
[...]

    | #(UNARY_MINUS e:expr) {o = il.unaryMinus(e);}
[...]
    ;

The respective generated TreeParser code looks like the following:

1  case UNARY_MINUS:
2   {
3        AST __t25 = _t;
4        AST tmp20_AST_in = (AST)_t;
5        match(_t,UNARY_MINUS);
6        _t = _t.getFirstChild();
7        e = _t==ASTNULL ? null : (AST)_t;
   // I'd like to avoid the call on line 8 since I'm calculating the
   // result myself on line 12
8        expr(_t); 
9        _t = _retTree;
10       _t = __t25;
11       _t = _t.getNextSibling();
12       o = il.unaryMinus(e);
13       break;
14  }

Questions:
1) I thought when I say "#(UNARY_MINUS e:expr)" instead of
"#(UNARY_MINUS e=expr)", ANTL would not bother calculating expr (since
the AST itself is requested, not the value of the expr). Is there a
way to tell ANTL that I will walk this part of the tree and calculate
the result myself?

2) Lines 9, 10, 11 seem a bit strange. Not sure I understand the
meaning of this.

Any ideas? Thanks




 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/