I've been working on a re-implementation of the Unicode Collation Algorithm for twitter-cldr-rb since the existing one is a bit old (2012-era) and Unicode has changed significantly since then. It's been a slow process, but I'm gradually piecing away at the conformance test failures. On Friday I was down to 3k (of 103k), last night it was 97, this morning 38, and as I write this in the evening 32! Progress!