Stabilize fixed-length slice patterns (excluding `..`)

[nikomatsakis]

This is a sub-issue of the larger tracking issue for slice patterns (#23121). It specifically tracks the plans to stabilize fixed-length slice patterns -- that is, excluding the .. operator (which is entangled with some syntactic questions). There is an implementation PR in #48516.

Summary

Using this feature, a pattern like [a, b, c] can be used to match against fixed-length arrays:

let arr = [1, 2, 3];
let [a, b, c] = arr; // where arr: [T;3]
let [a, ref b, c] = arr; // where arr: [T;3]

One can also use such a pattern to match against slices:

let arr = &[1, 2];
match arr {
    [a, b, c] => { /* this arm will not apply */ }
    [a, b] => { /* this arm will */ }
    _ => { /* this wildcard is required, since we don't know length statically */
}

Notable points and tests

Here are some of the things I think we should be testing. I didn't have time to find if we have all these tests, perhaps someone can help (cc @petrochenkov, @mikhail-m1 ?)

• test showing irrefutable matches (let [a, b] = ) applied to fixed-length to move things out
• test showing irrefutable matches (let [a, b] = ) applied to create reference
• test showing attempt to move from a borrowed array or slice gives error
  • move-out-of-slice1.rs sort of tests that, but for an oddly special case involving box patterns
• test that wildcard is required when matching slice with fixed-length arrays
• match-vec-fixed tests that we get warnings for two slice patterns of same length
• other cases?

Things not stabilized

• The main thing not stabilized is the ability to create a slice of unmatched items.

[nikomatsakis]

@rfcbot fcp merge

I propose we merge the subset above. @petrochenkov has a PR here:

#48516

I'd like to see us document the tests better, but I think that we have them =)

[rfcbot]

Team member @nikomatsakis has proposed to merge this. The next step is review by the rest of the tagged teams:

• @aturon
• @cramertj
• @eddyb
• @nikomatsakis
• @nrc
• @pnkfelix
• @withoutboats

No concerns currently listed.

Once a majority of reviewers approve (and none object), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

[petrochenkov]

Here are some of the things I think we should be testing. I didn't have time to find if we have all these tests, perhaps someone can help (cc @petrochenkov, @mikhail-m1 ?)

There are many tests testing slice patterns (at least 67 of them contain feature(slice_patters)), it's hard to find specific examples.
compile-fail/borrowck contains tests for moving out elements in array patterns erroneously.
match-vec-*/match-slice-* tests in compile-fail test various pattern properties like exhaustiveness and number of elements.
#47926 added tests for legally moving out elements from fixed-size arrays and dynamic drops.
irrefutable-slice-patterns.rs creates references with irrefutable slice patterns.

[rfcbot]

🔔 This is now entering its final comment period, as per the review above. 🔔

[nikomatsakis]

@petrochenkov good enough, thanks

[rfcbot]

The final comment period is now complete.