match_same_arms: suppress when reordering arms?

[birkenfeld]

match_same_arms will not consider the order of match arms.

Since it is specified that arms are checked in order, this leads to suggestions with two potential problems:

• less optimal performance, when expensive arm conditions are checked too early
• changes in semantics

Example for the latter:

fn main() {
    let x = match 2 {
        1 ... 3 => false,
        0 ... 5 => true,
        _ => false
    };
}

If this seems bad anyway because of overlap, consider the slightly more complex

fn main() {
    let y = 1;
    let x = match 2 {
        1 ... 3 => false,
        0 ... 5 if y == 2 => true,
        _ => false
    };
}

[oli-obk]

We can try to analyze whether order changes don't change semantics. A simple analysis would probably assume that only the order of different enum variants is irrelevant or the order of non overlapping int patterns

[Arnavion]

Duplicate of #860, though that one initially meandered into an unrelated discussion and this one has a more realistic example.