Suggest remove deref for type mismatch

[chenyukang]

Fixes #106496

[rustbot]

r? @compiler-errors

(rustbot has picked a reviewer for you, use r? to override)

[compiler-errors]

There are a few other instances of "consider removing the borrow" in this function that I think use span-to-snippet instead of looking at the HIR. Do you mind investigating those and migrating those too? It's alright if you don't want to.

[compiler-errors]

@rustbot author

Pls mark as ready-to-review if you don't want to apply the suggestion.

[chenyukang]

sure, I will take a look at it. Michael Goulet ***@***.***> 于 2023年1月23日周一 00:51写道：

…

There are a few other instances of "consider removing the borrow" in this function that I think use span-to-snippet instead of looking at the HIR. Do you mind investigating those and migrating those too? It's alright if you don't want to. — Reply to this email directly, view it on GitHub <#107203 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABYJ5USBH5S4GR2SEZJSHLWTVQSJANCNFSM6AAAAAAUDCK25I> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[chenyukang]

There are a few other instances of "consider removing the borrow" in this function that I think use span-to-snippet instead of looking at the HIR. Do you mind investigating those and migrating those too? It's alright if you don't want to.

I'm not quite understand your meaning, you mean this part?

rust/compiler/rustc_hir_typeck/src/demand.rs

Lines 1304 to 1341 in 2aa5555

	) if self.can_sub(self.param_env, checked, expected).is_ok() => {
	// We have `&T`, check if what was expected was `T`. If so,
	// we may want to suggest removing a `&`.
	if sm.is_imported(expr.span) {
	// Go through the spans from which this span was expanded,
	// and find the one that's pointing inside `sp`.
	//
	// E.g. for `&format!("")`, where we want the span to the
	// `format!()` invocation instead of its expansion.
	if let Some(call_span) =
	iter::successors(Some(expr.span), |s| s.parent_callsite())
	.find(|&s| sp.contains(s))
	&& sm.is_span_accessible(call_span)
	{
	return Some((
	sp.with_hi(call_span.lo()),
	"consider removing the borrow".to_string(),
	String::new(),
	Applicability::MachineApplicable,
	true,
	true
	));
	}
	return None;
	}
	if sp.contains(expr.span)
	&& sm.is_span_accessible(expr.span)
	{
	return Some((
	sp.with_hi(expr.span.lo()),
	"consider removing the borrow".to_string(),
	String::new(),
	Applicability::MachineApplicable,
	true,
	true,
	));
	}
	}

I go through the HIR usage of this function, seems it's necessary since use span-to-snippet to get source code and then search with pattern-matching is not correct in some cases.

rust/compiler/rustc_hir_typeck/src/demand.rs

Lines 1272 to 1286 in 2aa5555

	let prefix = match self.maybe_get_struct_pattern_shorthand_field(expr) {
	Some(ident) => format!("{ident}: "),
	None => String::new(),
	};
	if let Some(hir::Node::Expr(hir::Expr {
	kind: hir::ExprKind::Assign(..),
	..
	})) = self.tcx.hir().find_parent(expr.hir_id)
	{
	if mutability.is_mut() {
	// Suppressing this diagnostic, we'll properly print it in `check_expr_assign`
	return None;
	}
	}

[compiler-errors]

Oh, never mind. Ignore this then.

[compiler-errors]

I think this is fine to merge as-is then.

@bors r+

[bors]

📌 Commit 2aa5555 has been approved by compiler-errors

It is now in the queue for this repository.

[compiler-errors]

it would be nice to generalize this to multiple "&&" but whatever, this seems like a very common case already!