Justin Blank
Justin Blank
Cool! I haven't spent much time actively thinking about this since I created the initial issue, and I must admit, on second thought, I can't 100% reconstruct what I was...
The problem isn't gone, but some recent work to enable use of byteclasses is probably helpful. The regex `Holmes.{0,25}Watson|Watson.{0,25}Holmes` now generates a 340 KB class, instead of 2MB.
Subsequent updates to encoding make `Holmes.{0,25}Watson|Watson.{0,25}Holmes` only 245181 bytes.
Thanks for the note. I'll have a look at your benchmarks, and keep them in mind. Right now, I have a few things that I think need to be addressed...
@almondtools I was looking at the benchmarks--are there any scripts for handling the output?
Sorry, my earlier question was a bit vague. Yes, I was able to implement those in a branch I have locally, and doing so helped me find two bugs in...
Whoops, my apologies. I overlooked the command files.
Added unicode benchmarks in https://github.com/hyperpape/needle-benchmarks/blob/main/src/main/java/com/justinblank/strings/SherlockTranslatedUnicodeBenchmark.java, and the results are that the performance on unicode needles is quite bad, comparable to the standard library java regex or even worse in a...