> OCAML tooling is horrible

This is the old view. In 2020 OCaml tooling, if I may be brave to say so, is excellent. Check out the triumvirate of dune (build), opam (external dependencies) and ocaml-lsp (IDE smarts).

P.S. ocaml-lsp uses merlin in the backend (ignore that if you are not familiar with merlin).

I just spent 2 hours on this excellent tooling to try to upgrade a package with opam, I finally gave up after read all the Github issues and Stackoverflow, etc.

I spent two evenings (3 years ago, admittedly) to get the tooling setup so that I could install libraries. I got nowhere and gave up. I'm quite interested in the language but everything else seems to be a dumpsterfire.

I just tried it again, everything (m4, dune, opam, ocaml-lsp) installed just fine; then I created a simple hello world project (Quick start guide: https://dune.readthedocs.io/en/latest/quick-start.html) and I could compile and run it. Oh, how things appear to have changed.

You can use camomile library to handle unicode strings, count code points, perform locale aware lowercase / uppercase, etc

For utf8 regexes, you can use Pcre in ocaml, which is faster than camomile regexes.