NFA lexer construction is incredibly fast, but the execution is painfully slow. This is due to the closure algorithm being executed lots of times during the lexing. This could be avoided.
Create some tests that shows the poor performance of the runtime
Cache closure results for given state sets.
Probably improve the character range searches as well
NFA lexer construction is incredibly fast, but the execution is painfully slow. This is due to the closure algorithm being executed lots of times during the lexing. This could be avoided.
Create some tests that shows the poor performance of the runtime
Cache closure results for given state sets.
Probably improve the character range searches as well