Simon Krajewski

That sounds good to me. Any `suspend` function can implicitly suspend further calls, but if you do it from a "normal" function you have to be explicit.

Note that the "stand out" part doesn't necessarily mean that it has to be written in the code. As long as the compiler, and thus the IDE, know what suspends...

Isn't this just a matter of having an additional initial state in the state machine?

It's normal to go from general to specific. As long as we make sure we don't make decisions that block more specific "extensions" from following up, I don't really see...

Did we ever discuss what should happen for something like `0 ?? throw "error"` on static targets? Shouldn't that be a compiler error? The current implementation of `??` just always...

My initial thought here was that it's not surprising to see this not being optimized, but I'm very surprised that the version with the explicit `cast` works _better_ than the...

Yes, I already investigated that a bit and found that the `cf_expr` values were all `None`. This suggests some problem in the compilation server, not something eval-specific.

This is a top-down inference problem which we can reduce to this: ```haxe abstract A(String) { public function new() { this = ""; } @:arrayAccess public function arrayWrite(k:String, v:Array) {}...

Come to think of it, for consistency we would want this to work for the keys as well: ```haxe abstract A(String) { public function new() { this = ""; }...

The question has to be if the inner `@:meta` actually improves anything. If it just means adding it in front of every call then I don't see the point. However,...