Sébastien Doeraene

Phew, I got it minimized to ```scala class TestBody1 class TestBody2 class StartWithWord class EndWithWord class Matchers: extension (leftSideString: String) def should(body: TestBody1): Unit = () def should(body: TestBody2): Unit...

Don't hesitate to ping me when you would like me to take another look. I noticed you were still actively making changes, so I didn't pay attention to every push.

There are some test failures in the CI. Can you take a look?

Disclaimer: Comment written as an outsider user of Arcade. It doesn't represent the opinion of the maintainers in any way. Types provide *guarantees*. If Pyglet only says it accepts `int`s,...

~~This will need a rebase on top of #5192 once it's merged, in order to also change the calls to `toUint` introduced there. Otherwise ready for review.~~ Rebased.

I don't think so. It only affects new code combined with Java code. It cannot break anyone immediately. We should fix it, though, for sure, but not a blocker IMO.

> @sjrd what is your gut feeling on using weak references for invalidation tracking? It feels like we might get away with implicit invalidation by just throwing the cache away,...

Hah, that makes a lot more sense! Indeed that seems viable. We'll have to ES2021 when linking the JS linker, but that's acceptable if it makes the JVM linker better.

Ah I see. That makes sense. But wouldn't it be clearer to then list `"__dict__"` explicitly in the `__slots__`, rather than relying on the fact there happens to be a...

Those are fine implementations, but not as good as what's in this PR. In particular, caching the `max` is worthwhile, because divisions by non-constants are expensive. Otherwise it's pretty similar.