George Marques

For me the most intuitive behavior is that a local variable is only available after it is declared. Before that it is either an error (if it does not exist)...

> Obviously shadowing is a bad practice that we should not encourage. But should we disallow it? We already have a warning (which some users may not notice, but is...

I think it should not even try to call `parse_statement()` in this case. IMO the check should be in `parse_suite()` before doing the call to `parse_statement()`. The synchronization definitely can...

I've been thinking about this and I'm honestly not really sure. One one hand I understand why you have to do this, as it will otherwise share arrays that should...

I just remembered this PR: #56338 which essentially avoids the cases of sharing by making the constant map in the compiler use a different way of comparing values. It is...

> So you want this fix not to apply to usual dictionary keys? I don't think it is a problem for regular dictionary keys, it's probably even expected those are...

As I said before, it's okay if empty `Array[int]` is different than empty `Array[String]`. The problem is when comparing typed with untyped: ```gdscript var a: Array[int] = [1, 2, 3]...

It seems to me this still needs discussion on what the proper behavior should be. Only then we can say whether the implementation is optimal. There's also a change in...

I cannot reproduce this either, it was likely fixed in the meantime.

The GDScript analyzer should probably avoid the `preload` and leave loading to a later step. The problem is that then it does not know the type of the resource. It...