rustc: rename -Z emit-directives to -Z emit-artifact-notifications and simplify the output.

[eddyb]

This is my take on #60006 / #60419 (see #60006 (comment)).
I'm not too attached the "notifications" part, it's pretty much bikeshed material.
EDIT: for "artifact", @matklad pointed out Cargo already uses it (in #60464 (comment))

The first two commits are fixes that could be landed independently, especially the compiletest one, which removes the need for any of the normalization added in #60006 to land the test.

The last commit enables the emission for all outputs, which was my main suggestion for #60006, mostly to show that it's minimal and not really a "scope creep" (as suggested in #60006 (comment)).

cc @alexcrichton @nnethercote

[rust-highfive]

r? @pnkfelix

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

[matklad]

"compiler-artifact" is also used by cargo, so it's an ok name.

[alexcrichton]

I don't personally have a preference one way or another on this, so long as Cargo gets the notification for metadata seems reasonable to me!

[eddyb]

r? @alexcrichton

[alexcrichton]

Looks reasonable to me! There's some stray changes I don't know why are included though

[alexcrichton]

Where did these changes come from and/or how are they related to the PR's title?

[eddyb]

See the PR description - they remove the need for normalization, by fixing the compiletest bugs that affected #60006. IMO, the resulting tests are more clearly correct.

[alexcrichton]

Ah ok, sorry for missing that!

[alexcrichton]

@bors: r+

[bors]

📌 Commit fa9e8c829fdcfb08085a60e2d34367414b50a103 has been approved by alexcrichton

[bors]

⌛ Testing commit fa9e8c829fdcfb08085a60e2d34367414b50a103 with merge c940d25e110a9aabec4917bb262f1a6bc4ef9ed1...

[bors]

💔 Test failed - status-appveyor

[eddyb]

cc @davidtwco (#53196) looks like $TEST_BUILD_DIR doesn't work well on windows?

[davidtwco]

@eddyb I'll kick off a build of this on Windows and see if I can figure out what it's doing wrong. I probably won't get to that until tomorrow though.

From a quick glance, I notice that the $TEST_BUILD_DIR path is printed with .to_str().unwrap():

rust/src/tools/compiletest/src/runtest.rs

Line 3169 in a19cf18

normalized = normalized.replace(test_build_dir.to_str().unwrap(), "$TEST_BUILD_DIR");

Above that, the $SRC_DIR is printed with .display().to_string():

rust/src/tools/compiletest/src/runtest.rs

Line 3160 in a19cf18

src_dir.display().to_string()

I wonder if this would be as simple as changing the $TEST_BUILD_DIR and $BUILD_DIR paths to print the same way as $SRC_DIR?

[nnethercote]

It took me multiple goes to get the normalization right due to various platform-specific differences. I suspect you might get to go through the same experience I had. I suspect you'll have a wasm-related failure at some point.

[eddyb]

@nnethercote I fixed the underlying issue (-o being passed even when not building a binary to execute), which means I always get the right filename.
And I need no custom normalization, this is just another compiletest bug I want to fix instead of working around it.

[eddyb]

Got distracted looking at the history of UI test normalization, I'm thinking 0cfc173 can be partially or fully reverted if the problems in question were caused by this bug.

Anyway, in this case, the failure makes a lot of sense, because if json logic is missing for most path normalizations, and this is a JSON test, so the replacement from \ to \\ is essential.

[eddyb]

@bors r=alexcrichton

[bors]

📌 Commit c89a131 has been approved by alexcrichton

[bors]

⌛ Testing commit c89a131 with merge b8fa4cb...

[bors]

☀️ Test successful - checks-travis, status-appveyor
Approved by: alexcrichton
Pushing b8fa4cb to master...