Resolve visibility paths as modules not as types.

[cjgillot]

Asking for a resolution with opt_ns = Some(TypeNS) allows path resolution to look for type-relative paths, leaving unresolved segments behind. However, for visibility paths we really need to look for a module, so we need to pass opt_ns = None.

Fixes #109146

r? @petrochenkov

[petrochenkov]

CI is failing.
@rustbot author

[est31]

is there a test for what happens if the path is valid? newy is an invalid path but theoretically there could be a bug where if the path is valid but points to a struct or a function it accepts that.

[cjgillot]

@est31 there are additional tests in tests/ui/resolve/resolve-bad-visibility.rs

One surprising consequence of this PR is a different diagnostic between structs and enums. Enums are modules for resolve, so trigger E0577, while structs and unions are not, triggering E0433.

@rustbot ready

[petrochenkov]

@bors r+

[bors]

📌 Commit 69ea85d has been approved by petrochenkov

It is now in the queue for this repository.

[petrochenkov]

It fixes the ICE, but generally I don't think this is a right thing to do.

Paths in trait contexts for example, also do not support unresolved segments, but do not result in ICEs because they are only resolved during late resolution and late resolution uses enum PathSource and its methods like defer_to_typeck and is_expected instead of bare Namespace.

Right now None in resolve_path means an import prefix specifically, i.e. TypeNs + DefKind::{Mod,Enum,Trait} + defer_ty_typeck -> false restrictions.
You can see the consequences of that in tests/ui/use/use-self-type.stderr where the error message talks about imports now.

Ideally resolve_path should use something like PathSource too.
And resolve_visibility should use it as well, but with slightly different restrictions (TypeNs + DefKind::Mod + defer_ty_typeck -> false).

[bors]

⌛ Testing commit 69ea85d with merge 28b6607...

[bors]

☀️ Test successful - checks-actions
Approved by: petrochenkov
Pushing 28b6607 to master...

[rust-timer]

Finished benchmarking commit (28b6607): comparison URL.

Overall result: ❌✅ regressions and improvements - no action needed

@rustbot label: -perf-regression

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	1.3%	[1.3%, 1.3%]	1
Improvements ✅ (primary)	-0.3%	[-0.3%, -0.3%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.3%	[-0.3%, -0.3%]	1

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.7%	[2.7%, 2.7%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.2%	[-1.2%, -1.2%]	1
All ❌✅ (primary)	2.7%	[2.7%, 2.7%]	1

Cycles

This benchmark run did not return any relevant results for this metric.

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 648.817s -> 649.629s (0.13%)