[WIP] rustc_typeck: make "global" WF requirements hard errors, in type aliases.

[eddyb]

Based on #54033, a theory I want to test on crater.
Unlike #54033, this only errors on "unusable" aliases, as in, no choice of type parameters can make them WF, because they include nested non-WF types/constants.

r? @nikomatsakis

[eddyb]

@bors try

[bors]

⌛ Trying commit 1713f89 with merge f1a3f05968ed2a674e09b6efa4ad9df90816e2b6...

[rust-highfive]

The job x86_64-gnu-llvm-5.0 of your PR failed on Travis (raw log). Through arcane magic we have determined that the following fragments from the build log may contain information about the problem.

Click to expand the log.

[01:12:37] ....................................................................................................
[01:12:40] .......................................................i............................................
[01:12:43] ....................................................................................................
[01:12:46] ....................................................................................................
[01:12:48] ...iiiiiiiii........................................................................................
[01:12:55] ....................................................................................................
[01:12:58] ......................................................................................i.............
[01:13:01] ....................................................................................................
[01:13:04] .........................................i.i..ii....................................................
---
travis_time:start:test_rustdoc
Check compiletest suite=rustdoc mode=rustdoc (x86_64-unknown-linux-gnu -> x86_64-unknown-linux-gnu)
[01:25:29] 
[01:25:29] running 260 tests
[01:25:56] ...........F...........i............................................................................
[01:26:29] ............................................................
[01:26:29] failures:
[01:26:29] 
[01:26:29] ---- [rustdoc] rustdoc/const-evalutation-ice.rs stdout ----
[01:26:29] ---- [rustdoc] rustdoc/const-evalutation-ice.rs stdout ----
[01:26:29] 
[01:26:29] error: rustdoc failed!
[01:26:29] thread 'main' panicked at 'Some tests failed', tools/compiletest/src/main.rs:496:22
[01:26:29] status: exit code: 1
[01:26:29] command: "/checkout/obj/build/x86_64-unknown-linux-gnu/stage2/bin/rustdoc" "-L" "/checkout/obj/build/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib" "-L" "/checkout/obj/build/x86_64-unknown-linux-gnu/test/rustdoc/const-evalutation-ice/auxiliary" "-o" "/checkout/obj/build/x86_64-unknown-linux-gnu/test/rustdoc/const-evalutation-ice" "/checkout/src/test/rustdoc/const-evalutation-ice.rs"
[01:26:29] ------------------------------------------
[01:26:29] 
[01:26:29] ------------------------------------------
[01:26:29] stderr:

I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact @TimNN. (Feature Requests)

[bors]

☀️ Test successful - status-travis
State: approved= try=True

[eddyb]

@craterbot run start=master#f7f4c500b46603386e940f116b469c7adc043a6d end=try#f1a3f05968ed2a674e09b6efa4ad9df90816e2b6 mode=check-only

[craterbot]

👌 Experiment pr-54394 created and queued.

ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[craterbot]

🚧 Experiment pr-54394 is now running on agent aws-2.

ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[craterbot]

🎉 Experiment pr-54394 is completed!
📰 Open the full report.

⚠️ If you notice any spurious failure please add them to the blacklist!
ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[eddyb]

@nikomatsakis Can you take a look at these results? I'm confused because I expected to not see any errors - maybe the crater run should've been done with cap-lints=warn so #![deny(warnings)] doesn't make the warning an error, but does that explain everything?

EDIT: I looked at a random sample, and only a few were {deny,forbid}(warnings), and the remaining ones were either caused by nalgebra or diesel.
EDIT2: okay, it looks like both nalgebra and diesel only have errors with help: messages indicating they should be lints, but I hadn't changed the respective error logic to support lints.
EDIT3: @lqd helped with combing through the logs, and found a few intended regressions (most of them depend on domain, but there's overall not a lot, so we can do future-compat warnings here).

[XAMPPRocky]

Triage; @nikomatsakis Hello, are you able to review this PR?

[eddyb]

@Aaronepower This is not waiting on a review, I still have to finish the PR.

[nikomatsakis]

@eddyb

Can you take a look at these results? I'm confused because I expected to not see any errors

Just saw this, sorry...

[TimNN]

Ping from triage @eddyb: what is the status of this PR?

[nikomatsakis]

I'm going to close this PR for now. I do plan to start opening some PRs and issuing warnings, per the plan discussed in #55222