Fix documentation that disagrees with code

[fhartwig]

Minor fix: the comment currently disagrees with the lint declaration.

Out of curiosity: why is this lint allow by default? It seems to me like this should be uncontroversial.

[Manishearth]

I actually can't remember why it's Allow. @llogiq ?

[llogiq]

I remember that we had a major false positive with &mut Vec, where any of the functions &mut [_] doesn't have is used.

[Manishearth]

Oh, right.

[fhartwig]

My reading of the lint is that it only checks for shared references (and the tests support that reading), so I don't understand how this would cause false positives.

[Manishearth]

&mut Vec<T> has more capabilities than &mut [T], so there are cases when you may want to allow that.

Making the test ignore &mut Vec would be an acceptable change (and would make it okay to be Allow)

[fhartwig]

I understand that part, but the lint only warns about &Vec, not about &mut Vec, unless I'm missing something. The tests even say:

fn do_vec_mut(x: &mut Vec<i64>) { // no error here
    //Nothing here
}

[Manishearth]

Oh, hm, we do check for immutability.

I'm willing to accept a PR that makes this warn, then 😄