Use of RE2/J in JRuby!

[Charles Nutter]

Hello! I am excited to find your project!

I am a maintainer of JRuby. We have our own regular expression engine called Joni, which is a Java port of the Oniguruma regular expression engine. It works directly with byte[] and supports multiple RE grammars and string encodings.

https://github.com/jruby/joni

Recently, the folks who maintain the original CRuby and its copy of Oniguruma (now called Onigmo) have added a DFA for non-backtracking regular expressions, and it is now a "feature" that such expressions will have linear execution time. The original creator of Joni no longer works on the project, so we will have to port their DFA ourselves... or... maybe we can use RE2/J?

I am *extremely* happy to see that RE2/J can work directly with bytes; this is a major limiting factor for us, since Ruby string behavior requires that all JRuby strings be byte[]-based. I believe it will be possible for us to use RE2/J for the expressions it can handle, falling back on our full backtracking engine in Joni otherwise.

I'd also love to collaborate on this. Perhaps RE2/J can be evolved to support other character encodings? If it could, then we could integrate it directly into Joni, to have a DFA when possible, but backtracking when needed.

In any case, thanks for this library, and I'm happy to chat about possibilities here or on the JRuby Matrix channel.