Fix invalid while_let_on_iterator suggestion

[jfirebaugh]

Slice patterns without .. are refutable. (Well, except if the match source is an array type -- does is_refutable have enough information to check this though?)

Fixes #3780.

[oli-obk]

It might still be irrefutable if head or tail are irrefutable

[jfirebaugh]

Are you referring to matches on array types? As @mikerite points out, the only irrefutable slice pattern on slices is [..].

[oli-obk]

does is_refutable have enough information to check this though?)

Not sure, but we can probably pass in the relevant information

[ghost]

The patterns [a,..] and [..,b] require at least 1 element so they won't match []. This makes me think that the only irrefutable slice pattern is [..] so maybe just check for that.

[jfirebaugh]

does is_refutable have enough information to check this though?

It appears so; I've added type-checking in the latest commit, though I am not too sure of its correctness. I see that MIR handles the case where the pattern is ty::Ref, as well as ty::Error. Does anyone know in what situations these arise?

As an aside, it would be nice to lean on the compiler for something like is_refutable. What is the stance toward making items public in, say, rustc_mir solely for the benefit of external tooling like clippy? For example, something like this would presumably be more robust than clippy's current implementation.

[bors]

☔ The latest upstream changes (presumably #3803) made this pull request unmergeable. Please resolve the merge conflicts.

[flip1995]

Ping @jfirebaugh. I'm going over old PRs, that were abandoned by us reviewers (sorry for that!) or by the authors. Are you still interested in completing this?

[flip1995]

Ping from triage @jfirebaugh. Sorry for neglecting this PR and thanks for your contribution. If you want to continue working on this, feel free to reopen and ping me. I'll be happy to help you wrapping this up.

[hkalbasi]

finished in #5559
@rustbot label -S-inactive-closed