Herwin

I think only showing `**{` when it's empty would be clearer, the `foo.bar(**{:x => 1})` looks very complex, and it doesn't really add anything

> Update Unicode to Version 15.0.0 and Emoji Version 15.0. [[Feature #18639](https://bugs.ruby-lang.org/issues/18639)] It looks like that one has been covered in commit 8caf02ddc1d951425a5eea35fe93d5a1492f9c3a

> Introduce `IO.new(..., path:)` and promote `File#path` to `IO#path`. > [[Feature #19036](https://bugs.ruby-lang.org/issues/19036)] This looks to be covered in commit 62c1bbe88028961acd8fc059c3d14c24a95e0316

Snippet of the observed behaviour (Ruby 3.1) ``` irb(main):001:0> io = IO.new(1, 'w:ISO-8859-1') => # irb(main):002:0> io.internal_encoding => nil irb(main):003:0> io.external_encoding => # irb(main):004:0> io = IO.new(1, 'w:ISO-8859-1', internal_encoding: 'ISO-8859-1')...

For what it's worth: https://github.com/natalie-lang/natalie/pull/779 has a bit of a write up with the issues I encountered

The Ci uses G++ 7.5, which is kind of ancient. From the top of my head, you need version 8 to get full support for C++-17, which introduced the `filesystem`...

I tried compiling the parent of that commit, but that didn't solve the issue, guess it's not completely the same issue.

Somewhere between commit ea03e122 and commit 88eb2d83 it has been changed from a segfault into an illegal exception. That first commit is the last one of the original source.

Some more experiments: * Adding some unused fields to the `jt` structure before the `mfree` struct, or as a first element in the `mfree` struct does not change the behaviour,...

As a side note: I'm pretty much using this ticket as a whiteboard, subscribing to changes of it might get a bit spammy.