Fix runnable detection for #[tokio::test]

[HKalbasi]

fix #15141

It is hacky, and it wouldn't work for e.g. this case:

use ::core::prelude;

#[prelude::v1::test]
fn foo() {
}

But it works for the tokio case. We should use the name resolution here somehow, and after that we should probably also get rid of the ast based test_related_attribute function.

[Veykril]

Why are we changing the is_test impl, can't we just bring back the use of test_related_attribute where it was removed?

We could also try a different idea. Fetch the source of the function (or whatever def), walk up the macro expansions its in and check if any of those comes from the test attribute. If yes, it's a test runnable. Like

rust-analyzer/crates/hir-expand/src/lib.rs

Lines 343 to 352 in 7c8ae35

	/// Return whether this file is an include macro
	pub fn is_include_macro(&self, db: &dyn db::ExpandDatabase) -> bool {
	match self.macro_file() {
	Some(macro_file) => {
	let loc: MacroCallLoc = db.lookup_intern_macro_call(macro_file.macro_call_id);
	loc.def.is_include()
	}
	_ => false,
	}
	}

[HKalbasi]

Why are we changing the is_test impl, can't we just bring back the use of test_related_attribute where it was removed?

I would like to get rid of the test_related_attribute function in future. It detects any proc macro with test in the start or end of name if I understand it correctly, which is bad. It looks like a pre macro expansion thing to me.

We could also try a different idea. Fetch the source of the function (or whatever def), walk up the macro expansions its in and check if any of those comes from the test attribute. If yes, it's a test runnable.

I guess we don't do macro expansion for the test attribute, do we? If we do, we can just check for #[rustc_test_maker] or emit some fictional attribute with a unique name and check for that.

[Veykril]

Ah right, we are skipping the test/bench expansion as its just wasting memory.

[HKalbasi]

Ah right, we are skipping the test/bench expansion as its just wasting memory.

Aha, that's reasonable. Is there a way to do name resolution on attributes without doing relying on macro expansion?

[HKalbasi]

I'm going to merge this to fix the tokio issue sooner, so that we can procrastinate think about the ideal solution without pressure.

@bors r+

[bors]

📌 Commit 97c48f9 has been approved by HKalbasi

It is now in the queue for this repository.

[bors]

⌛ Testing commit 97c48f9 with merge 5d4dd9f...

[bors]

💔 Test failed - checks-actions

[HKalbasi]

@bors retry

[bors]

⌛ Testing commit 97c48f9 with merge 729f5d0...

[bors]

💔 Test failed - checks-actions

[HKalbasi]

@bors r+

[bors]

📌 Commit 7901538 has been approved by HKalbasi

It is now in the queue for this repository.

[bors]

⌛ Testing commit 7901538 with merge ad434fc...

[bors]

☀️ Test successful - checks-actions
Approved by: HKalbasi
Pushing ad434fc to master...

[bors]

☀️ Test successful - checks-actions
Approved by: HKalbasi
Pushing ad434fc to master...

[bors]

👀 Test was successful, but fast-forwarding failed: 422 Changes must be made through a pull request.