Token - ASTNode relationship

[Tkrauss]

We need the support to associate tokens with an ASTNode for the purpose of debugging. I propose two alternatives, each associated with one method of the ASTNode Option 1:

// this method returns a interval-like object. This means, every token between the token tuple.left and the token tuple.right ( both inclusive) is in this subtree.
Tuple<Token,Token> coverage()

Option 2:

// this method returns the list of tokens, whose are covered within this ast-node ( and its complete subtree). 
// node1.coverage() is real subset of node2.coverage() <=> node1 is in subtree of node2
// Plus, the returned list must be ordered in a left-to-right-fashion.
List<Token> coverage()

I prefer option 2, since it is easy implementable using the current iterator ( at least, in the javabite-implementation) and ther's no need for more classes. What do you think?

[leinge]

I'd prefer the option 2, too! But I can't see why one node should hold all the tokens of its subtree. This seems very redundant to me, and I can't imagine a situation where this would be necessary atm...?

[Tkrauss]

In theory, this would be the choice of the implementation. Since we need this method only if something went wrong and additional just for a (probably small) subtree of the whole AST is the subject, there's no need to hold the value in each node in every situation. I would prefer an ad-hoc computation, but -- like i already said -- this is a matter of the implementation.

[Tkrauss]

So... if there are no more pleas, i'll define the interface as it was described (option 2) at 8pm today.