RFC: Irrefutable patterns in closure argument lists

[bstrie]

Pre-demoding, this sort of thing is valid:

fn main() {
    for ["Alice", "Bob", "Carol"].each |name| {
        do task::spawn {
            let v = rand::Rng().shuffle([1, 2, 3]);
            for v.each |num| {
                io::print(fmt!("%s says: '%d'\n", name, num))
            }
        }
    }
}

This looks very nice.

But post-demoding, it looks like this:

fn main() {
    for ["Alice", "Bob", "Carol"].each |name| {
        let name = *name;  // because `name` is now a pointer
        do task::spawn {
            let v = rand::Rng().shuffle([1, 2, 3]);
            for v.each |num| {
                io::print(fmt!("%s says: '%d'\n", name, *num))  // `num` is also a pointer
            }
        }
    }
}

Compared to the first example, this looks quite poor .

@nikomatsakis has mentioned (here) that if Rust supported irrefutable patterns in argument lists, the example would instead become:

fn main() {
    for ["Alice", "Bob", "Carol"].each |&name| {
        do task::spawn {
            let v = rand::Rng().shuffle([1, 2, 3]);
            for v.each |&num| {
                io::print(fmt!("%s says: '%d'\n", name, num))
            }
        }
    }
}

Which looks rather nice.

The inability to do this has already resulted in #3550 as well as the loss of one very nice front-page code example. I can foresee a lot of grumbling over the lack of this feature.

[nikomatsakis]

I thought there was an issue about this, but I can't find it. I am strongly in favor of permitting irrefutable patterns in argument lists (not just on closures). I think there is no opposition.

[eholk]

+1

[bstrie]

I know there's been some work towards this, how much is left to be done?

[bstrie]

Fixed by 0fc9523.