Nice to finally have this (assuming it goes through) but it's still not a general concept of immutability. It's only being applied to dicts and sets, which have immutable counterparts (i.e., to built-in types that are paired up as a mutable and immutable version of the same thing, excluding lists and tuples which are of course already handled). I can't see it leading to any kind of compile-time reasoning about mutability, which is unfortunate.