[antlr-interest] Weird problem with semantic predicates. Incorrect warning message

[Gavin Lambert]

At 09:51 1/05/2009, Gabriel Petrovay wrote:
 >ANTLR could either ignore ambiguous rules (if the computed k is
 >toooo large, give up) and proceed with normal computation.
 >Another thing ANTLR could do is be to compute k per-rule and
 >not per grammar.

k is already computed per-rule, or possibly even per-alt (in v3; 
in v2 you had to specify it explicitly).  But in cases where ANTLR 
can't figure it out it will default to k=1, since that gives the 
best performance (if not always the correct result, when the 
grammar is ambiguous).

It always helps to try to remove ambiguity and to keep your 
grammar as left-factored as possible.  This will not only make 
ANTLR's job easier, it will also improve runtime performance :)