Jeremy Evans

This needs more work to handle `{**nil}`, `f(a: 1, **nil)`, and other cases. I'll work on that later today.

> I benchmarked this PR with headline benchmarks of yjit-bench: > > ## Interpreter > It speeds up psych-load and sequel, but it slows down activerecord. Is the slowdown intended?...

> @jeremyevans are you intending to merge this PR in for 3.3? I'm a wary of merging this just 10 days away from the release. No, this is not intended...

Agreed. In this case, I did not know prism would break. This PR was last rebased on January 8, and the prism CI that broke wasn't added until January 10....

In general, `rdoc-ref` is useful because it makes the references internally consistent. If all references are to URLs, then there is no need for `rdoc-ref`, but if you are referencing...

@wpolicarpo If you have time, can you clarify what about the PR made it unable to merge? I'm not sure about the test, but the `client.c` change looks very straight...

> Anyway, I'm not sure why we're having this conversation on the merits of early hints. This ticket is about whether we want a spec for it in Rack, it...

In general, test suites for the Ruby standard library do not document what is being tested to this level. That could be considered a good thing (less noise when updating...

I don't think we should backport this to either Rack 3.0 or 2.2. We should only be backporting security fixes to Rack 2.2, and only bug and security fixes to...

I think we should merge this. However, as it is a new feature and not a bug fix, I don't think we should backport it.