Shrink SelectionError a lot

[Noratrieb]

SelectionError used to be 80 bytes (on 64 bit). That's quite big. Especially because the selection cache contained `Result<_, SelectionError>. The Ok type is only 32 bytes, so the 80 bytes significantly inflate the size of the cache.

Most variants of the SelectionError seem to be hard errors, only Unimplemented shows up in practice (for cranelift-codegen, it occupies 23.4% of all cache entries). We can just box away the biggest variant, OutputTypeParameterMismatch, to get the size down to 16 bytes, well within the size of the Ok type inside the cache.

[rustbot]

r? @petrochenkov

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

[rustbot]

Some changes occurred to the core trait solver

cc @rust-lang/initiative-trait-system-refactor

[compiler-errors]

@bors try @rust-timer queue

[bors]

⌛ Trying commit b9b27667a60d6085e3f0d5f5ee765f1c9ccb3e65 with merge 15cc844067a9ed1563f8c11342cbbb1316a30f5c...

[bors]

☀️ Try build successful - checks-actions
Build commit: 15cc844067a9ed1563f8c11342cbbb1316a30f5c (15cc844067a9ed1563f8c11342cbbb1316a30f5c)

[rust-timer]

Finished benchmarking commit (15cc844067a9ed1563f8c11342cbbb1316a30f5c): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

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

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)	2.8%	[2.3%, 3.2%]	2
Regressions ❌ (secondary)	2.8%	[2.5%, 3.0%]	2
Improvements ✅ (primary)	-1.6%	[-2.5%, -0.5%]	10
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.8%	[-2.5%, 3.2%]	12

Cycles

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

[Noratrieb]

perf was neutral except for max RSS, which went down on diesel, makes sense. Given the small impact I can change this to the "light" version, just boxing away the large variant instead of this clever packing here.

[petrochenkov]

Left a couple of style comments.
@rustbot author

[Noratrieb]

Changed it to the less invasive variant.
@rustbot ready

[Noratrieb]

This looks pretty wrong, except it's not.

[compiler-errors]

expected_trait_ref in this fn is a misnomer. it comes from taking the self type from an obligation like fn(): Fn(i32) and turning it into the Fn trait that we know it does implement -- in this case, fn() implements Fn() from its signature. Then we equate fn(): Fn() ("expected") and fn(): Fn(i32) ("obligation"). But when we're reporting an error, the obligation trait ref is what we want to say was expected from the user's perspective.

we should probably rename the expected_trait_ref parameter in this fn.

[Noratrieb]

I renamed it to self_ty_trait_ref, not sure whether that's the best name.

[bors]

💔 Test failed - checks-actions

[petrochenkov]

@bors retry

[rust-log-analyzer]

A job failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

[Noratrieb]

@bors treeclosed=50
let's give GitHub some time to recover

[bors]

⌛ Testing commit e8ab648 with merge f8d8ffa...

[bors]

☀️ Test successful - checks-actions
Approved by: petrochenkov
Pushing f8d8ffa to master...

[rust-timer]

Finished benchmarking commit (f8d8ffa): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

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

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)	4.2%	[4.2%, 4.2%]	1
Regressions ❌ (secondary)	2.0%	[1.6%, 2.5%]	2
Improvements ✅ (primary)	-1.6%	[-2.4%, -0.5%]	9
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-1.0%	[-2.4%, 4.2%]	10

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: 660.266s -> 659.801s (-0.07%)