Using unicode literals as indentifier produces a weird warning about snake case

[eiennohito]

Sample code (uses Hiragana を as an identifier)

fn を(i: i32) -> i32 {
  i + 5
}

A warning:

test.rs:8:1: 10:2 warning: function `を` should have a snake case name such as `を`, #[warn(non_snake_case)] on by default
test.rs:8 fn を(i: i32) -> i32 {
test.rs:9   i + 5
test.rs:10 }

[nagisa]

The reason is that some scripts in Unicode (such as kana) have no associated case to them (i.e. they are no-case) and the lint only allows lowercase codepoints.

This is E-easy material, but it is unclear whether the lint should just outright accept any non-uppercase codepoint or we should refine the error message to not have “such as を” suggestion if the lint is triggered by no-case codepoints.

EDIT: I recall some other programming language compiler opting to keep emiting a warning if no-case codepoints are present in snake_case identifiers, but I don’t quite recall which language it was.