chore: multipart suggestions for let_unit_value lint

[scottgerring]

This should address #13099 for the let_unit test.

changelog: [let_unit]: Updated let_unit to use multipart_suggestions where appropriate

[rustbot]

r? @llogiq

rustbot has assigned @llogiq.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[scottgerring]

Flip1995 has the context of these, so if they have time:

r? flip1995

(But this is still a draft!)

[flip1995]

Great work! Just 2 codestyle comments.

Please also rebase your PR on top of the current master. We changed our CI recently and GitHub isn't good at picking this up on older PRs.

[scottgerring]

Hey @flip1995 thanks for the review! I rebased and addressed the resulting dogfooding lints and your own comments, so there's a bit of a delta on top of what you reviewed. Should be good to go, though, I reckon.

I am assuming you squash onto master, but if not, I can clean my commit history up in future 🤣

[flip1995]

Thanks! We don't squash on merge. So, yes, please clean up the commit history :)

[flip1995]

That's a bit of overkill 😅 I would suggest to change this to:

if !suggestions.is_empty() {
    let message = if suggestions.len() == 1 {
        "omit the `let` binding"
    } else {
        "omit the `let` binding and replace variable usages with `()`"
    };
    diag.multipart_suggestion(...);
}

[scottgerring]

Yep, that's way less awful. Let me clean it up 👍

[scottgerring]

@flip1995 that should be it! 🙏

[scottgerring]

Hey @flip1995 , do you have a chance to have a quick look at this? We're very close to finishing the whole issue!

[flip1995]

Sorry for not replying. I won't get to review any Clippy PRs until around christmas... I'll reassign

r? clippy

[blyxyas]

This line is completely removed, is this intended?

[blyxyas]

And sorry for the late review, I've opened this PR maybe 5 times and never got around to clicking that "Submit review" button

[scottgerring]

Hey @blyxyas no worries!

I think it is unnecessary now; this part here collects usages:

rust-clippy/clippy_lints/src/unit_types/let_unit_value.rs

Lines 90 to 92 in ec24388

	// Collect variable usages
	let mut visitor = UnitVariableCollector::new(binding_hir_id);
	walk_body(&mut visitor, body);

... and we then emit a lint based on whether or not we had any:

rust-clippy/clippy_lints/src/unit_types/let_unit_value.rs

Lines 99 to 105 in ec24388

	if !suggestions.is_empty() {
	let message = if suggestions.len() == 1 {
	"omit the `let` binding"
	} else {
	"omit the `let` binding and replace variable usages with `()`"
	};
	diag.multipart_suggestion(message, suggestions, Applicability::MachineApplicable);

It could be added back as a sort of "performance optimization guard" to remove the need to call walk_body, assuming that is_local_used is quicker, but it doesn't change the behaviour (I also double checked this!)

[blyxyas]

I've been all morning trying to measure this but it's in a very awkward spot to get a hold of how many heavy-lifting that function does.

Being that in the case that the body is visited is already in the linting-case, let's just ignore it.

[blyxyas]

LGTM, thanks! ❤️

[blyxyas]

I've been all morning trying to measure this but it's in a very awkward spot to get a hold of how many heavy-lifting that function does.

Being that in the case that the body is visited is already in the linting-case, let's just ignore it.