[antlr-interest] The power of backtracking in ANTLR

[Søren Andersen]

Does anyone have any detailed information on backtrack=true in ANTLR grammars?

I'm particularly interested in if it has some "magic" properties, other than merely allowing me to *not* refactor my grammar?
Asked another way - given a valid grammar using backtracking, can I *always* refactor it to no longer need backtracking, preserving the semantics and the fact that the grammar is LL(*)?
Or - can backtracking remove my grammar from LL(*)?

Hope my question is clear! (And not too stupid :)

Regards,

Søren Andersen