2 // Providing eof_ok information is problematic for this grammar
3 // because if we've seen "XZ" the stack looks like this:
5 // +---------------------+
13 // +---------------------+
20 // +---------------------+
29 // +---------------------+
31 // The problem is that to know that we are in an eof_ok state,
33 // 1. recognize that the IntFA is in a final state (no prob -- we do this already)
34 // 2. speculatively assume the token is finished
35 // 3. speculatively transition the GLA to its final state
36 // 4. speculatively take the x2 transition in rule a
37 // 5. speculatively take the z transition in rule a
39 // Of course this is possible, but it's complicated both in terms of programming
40 // effort and run-time efficiency.
46 a -> .x1=x y | .x2=x z;