Gate the type length limit check behind a nightly flag

[compiler-errors]

Effectively disables the type length limit by introducing a -Zenforce-type-length-limit which defaults to false, since making the length limit actually be enforced ended up having a worse fallout than expected. We still keep the code around, but the type length limit attr is now a noop (except for its usage in some diagnostics code?).

r? @lcnr -- up to you to decide what team consensus we need here since this reverses an FCP decision.

Reopens #125460 (if we decide to reopen it or keep it closed)
Effectively reverses the decision FCP'd in #125507
Closes #127346

[jackh726]

imo this is the right choice - there is more fallout than I think we anticipated with re-adding the type length limit.

r=me, but maybe we want to wait a day or so for anyone in the team to dissent

[compiler-errors]

@bors r=jackh726

we can easily turn this back on

[bors]

📌 Commit 938ed36 has been approved by jackh726

It is now in the queue for this repository.

[compiler-errors]

@bors rollup=never

[bors]

⌛ Testing commit 938ed36 with merge 88fa119...

[rust-highfive]

📣 Toolstate changed by #127670!

Tested on commit 88fa119.
Direct link to PR: #127670

💔 reference on windows: test-pass → test-fail (cc @Havvy @matthewjasper @ehuss).
💔 reference on linux: test-pass → test-fail (cc @Havvy @matthewjasper @ehuss).

[bors]

☀️ Test successful - checks-actions
Approved by: jackh726
Pushing 88fa119 to master...

[ehuss]

@compiler-errors Can you please update the reference at https://github.com/rust-lang/reference/blob/master/src/attributes/limits.md#the-type_length_limit-attribute? This PR has broken the test there, and I'm not sure how the documentation should be updated. Is this attribute deprecated now?

[compiler-errors]

@ehuss:

Sure. The attribute is not deprecated per se, but it doesn't do anything right now. Previously it didn't do anything in real code, but it was easier to make a pathological example like rust-lang/reference#1026 did after it previously broke in #72412. See #125460.

I'll just remove the test. Not sure what we should do with the documentation. I'll probably just mention the attr doesn't do anything (which isn't totally true -- it actually is used in diagnostics, but I don't want to encourage people to use the attr).

[rust-timer]

Finished benchmarking commit (88fa119): comparison URL.

Overall result: ✅ 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)	-	-	0
Improvements ✅ (primary)	-0.6%	[-1.4%, -0.2%]	28
Improvements ✅ (secondary)	-1.1%	[-1.5%, -0.4%]	8
All ❌✅ (primary)	-0.6%	[-1.4%, -0.2%]	28

Max RSS (memory usage)

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

Cycles

Results (primary -1.0%)

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)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-1.0%	[-1.1%, -0.8%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-1.0%	[-1.1%, -0.8%]	2

Binary size

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

Bootstrap: 702.998s -> 699.561s (-0.49%)
Artifact size: 328.72 MiB -> 328.65 MiB (-0.02%)