cc @tekknolagi

Sorry: new to ruby contributing. Should I force push or will you squash merge at the end?

I can squash at the end. If you have multiple logical commits that you would like preserved, you should force push and I can rebase

I can squash at the end. If you have multiple logical commits that you would like preserved, you should force push and I can rebase

Nope I don't. Please squash and overwrite the commit message with your own at the end. Thanks!

If you have time, please add some tests for this, but it's not dire. I will otherwise merge tonight or Monday

I would add tests, but I'm not too sure how to at the moment. We'd basically need to pass a malformed iseq to the translation function and have it fail on validation right? Not sure how to construct an iseq in Rust in the first place considering rb_iseq_t is an opaque struct in Rust. There's probably some way to do it in C land, but I'm not familiar enough with the Ruby internals to achieve that yet.

No, we'd construct the HIR manually---not from an ISEQ

Ah ok, I was under the impression you wanted an integration test for iseq_to_hir with the validator. That makes more sense now :). I'll add tests for this in another PR if you're okay with it. The fact it passes everything is a pretty strong indicator it's at least not detecting any false positives.

Oh nevermind, I just found out rpo_tests exists. I can just copy some tests from there. Will have it done by tomorrow.

The failures look to be flakiness of Ractor tests. This PR doesnt touch Ractors at all so I'm not concerned.