Peter Blackson

My algorithm has three passes, like Jeffrey Kegler's. I am eliminating nullable symbols. I added some highly effective optimizations with a binary heap. Besides that, it is similar to Jeffrey...

Do the YAEP/YACC measurements include the definition and processing of a grammar, that is e.g. building the LALR(1) tables? My benchmarks include processing of the grammar.

Oh okay. The similar times and memory usages make a lot of sense. I will try to figure out why YAEP was so slow in my runs. To optimize, I...

I didn't mean to say you use LA(LR)(k), I meant to say YACC does. In case I incorporated LL(1) into my parser, would you consider it faster, or would you...

My attempts at optimization failed so far. I would like to optimize by caching all computations for an Earley set. I will hash all Earley sets individually and compare hashes...

Thank you. I'm currently implementing this new optimization with set hashing and a Patricia trie. I found another small optimization: Earley items can be shrunk by 1/3 by tying the...

What you call distances, I call item origins after Jeffrey Kegler. I found that distances can be removed with very little complication for the algorithm. I found that I can...

New measurements show that 610 out of 12600 Earley sets can be cached, on the C benchmark. The cache is huge, so probably not worth using. I will try to...

My benchmarks include processing the grammar. New measurements were wrong, I can skip processing for 7000 out of 8500 Earley items. I'm having some issues with the Patricia trie implementation,...

I voulnteer to implement this. First steps added in #538