Rie

One alternative approach I could imagine might be preferable is to change the behavior of `DirAccess::make_default`, but I think that would require more substantial changes and I'm not sure I...

> Similar problems also exist on Android, because on Android, "res://" is also a real local directory. Is there an issue documenting the problem on Android? I'm not sure my...

> #87274 Although I didn’t mention Android in the introduction, I did initially discover this problem on Android. Ah yes, that does appear to be the same problem and my...

I just realized there is a problem in my code: After loading a resource pack, you can only see files that existed in res:// at the time the first pack...

> It feels like the current fix is the usable version, `res://` is not supposed to be modified at runtime, so synchronizing only the first time is good enough. I...

The last thing I need clarity on is whether and how to expose `PackedSourceDirectory` to users. At the moment, due to how I implemented the fix, users can call `load_resource_pack("res://")`...

> I don't think there is any need to expose `PackedSourceDirectory` at the moment. Let this PR only fix bugs. 4 can be considered for completion in subsequent PRs. Makes...

By doing some more tests I realized setting the argument `replace_files` to `false` does nothing when loading the first pack, even on the master branch. It is documented here: #89299....

As discussed earlier, I added a line to the documentation of `load_resource_pack` explaining the remaining limitation to `DirAccess`. The wording may be a bit awkward, I'd appreciate any feedback if...

I'm failing a unit test for the macOS editor. Unfortunately I don't have access to any Apple computers... does anyone have an idea for what could be causing this? I...