Relate alias ty with variance #116219

compiler-errors · 2023-09-28T06:41:53Z

In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀.

This change is a bit more invasive, but I'd rather not special-case it here and then have it break elsewhere. I'm doing a perf run to see if the extra call to def_kind is that expensive, if it is, I'll reconsider.

r? @lcnr

compiler-errors · 2023-09-28T06:42:00Z

@bors try @rust-timer queue

bors · 2023-09-28T06:43:09Z

⌛ Trying commit e32b697 with merge 484b61b...

…-variance, r=<try> Relate alias ty with variance In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀. This change is a bit more invasive, but I'd rather not special-case it [here](https://github.com/rust-lang/rust/blob/aeaa5c30e5c9041264a2e8314b68ad84c2dc3169/compiler/rustc_trait_selection/src/solve/alias_relate.rs#L171-L190) and then have it break elsewhere. r? `@lcnr`

compiler-errors · 2023-09-28T06:45:47Z

@bors try

bors · 2023-09-28T06:45:47Z

⌛ Trying commit f310e3c with merge 834904f...

…-variance, r=<try> Relate alias ty with variance In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀. This change is a bit more invasive, but I'd rather not special-case it [here](https://github.com/rust-lang/rust/blob/aeaa5c30e5c9041264a2e8314b68ad84c2dc3169/compiler/rustc_trait_selection/src/solve/alias_relate.rs#L171-L190) and then have it break elsewhere. r? `@lcnr`

bors · 2023-09-28T06:46:54Z

⌛ Trying commit f310e3c with merge 2e44f1a...

…-variance, r=<try> Relate alias ty with variance In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀. This change is a bit more invasive, but I'd rather not special-case it [here](https://github.com/rust-lang/rust/blob/aeaa5c30e5c9041264a2e8314b68ad84c2dc3169/compiler/rustc_trait_selection/src/solve/alias_relate.rs#L171-L190) and then have it break elsewhere. r? `@lcnr`

fmease · 2023-09-28T07:01:32Z

compiler/rustc_middle/src/ty/relate.rs

+                    b.args,
+                    false, // do not fetch `type_of(a_def_id)`, as it will cause a cycle
+                )?,
+                DefKind::AssocTy | DefKind::AssocConst | DefKind::TyAlias => {


What about lazy type aliases? Shouldn't they be related with variances, too?

I considered this, but it shouldn't matter in practice, since they're always normalizable.

This only really concerns things that must be considered as aliases.

bors · 2023-09-28T07:56:54Z

☀️ Try build successful - checks-actions
Build commit: 2e44f1a (2e44f1a9b70ccff3d894e6dc190b49b3580650de)

rust-timer · 2023-09-28T10:55:01Z

Finished benchmarking commit (2e44f1a): comparison URL.

Overall result: ❌✅ regressions and improvements - 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.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +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)	0.2%	[0.2%, 0.2%]	2
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.3%	[-0.3%, -0.3%]	1
Improvements ✅ (secondary)	-0.3%	[-0.3%, -0.3%]	1
All ❌✅ (primary)	0.0%	[-0.3%, 0.2%]	3

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)	-	-	0
Regressions ❌ (secondary)	2.6%	[2.6%, 2.6%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.1%	[-1.1%, -1.1%]	2
All ❌✅ (primary)	-	-	0

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: 631.947s -> 631.255s (-0.11%)
Artifact size: 317.35 MiB -> 317.36 MiB (0.00%)

lcnr · 2023-09-28T12:57:47Z

compiler/rustc_middle/src/ty/relate.rs

@@ -273,7 +274,20 @@ impl<'tcx> Relate<'tcx> for ty::AliasTy<'tcx> {
        if a.def_id != b.def_id {
            Err(TypeError::ProjectionMismatched(expected_found(relation, a.def_id, b.def_id)))
        } else {
-            let args = relation.relate(a.args, b.args)?;
+            let args = match relation.tcx().def_kind(a.def_id) {
+                DefKind::OpaqueTy => relate_args_with_variances(


shouldn't it be fine to use variances_of for all other aliases as well? at this point you could use relation.relate_item_args which avoids any lookup in Equate (and needs to call variances_of regardless if we're not in Equate)

I don't think variances_of is impl'd for projections. I guess I could impl it, but that may have other perf implications.

hmm, alternatively, instead of using the DefKind, have the Ty::relate provide the AliasKind to the substs relate.

do we need ProjectionTy to implement Relate itself, or would an inherent method relate which also takes the AliasKind be possible?

do we need ProjectionTy to implement Relate itself, or would an inherent method relate which also takes the AliasKind be possible?

This is pretty difficult, because in order to invoke a relation, you need something that implements Self: ToTrace. That requires Self: Relate. I guess I could poke a hole through with something like trace, but even then, it doesn't seem really ergonomic. Or I could separate out ToTrace from Relate or something...

This also makes this code significantly uglier:

rust/compiler/rustc_trait_selection/src/solve/alias_relate.rs

Lines 171 to 190 in 56ada88

fn assemble_subst_relate_candidate(

&mut self,

param_env: ty::ParamEnv<'tcx>,

alias_lhs: ty::AliasTy<'tcx>,

alias_rhs: ty::AliasTy<'tcx>,

direction: ty::AliasRelationDirection,

) -> QueryResult<'tcx> {

self.probe_misc_candidate("args relate").enter(|ecx| {

match direction {

ty::AliasRelationDirection::Equate => {

ecx.eq(param_env, alias_lhs, alias_rhs)?;

}

ty::AliasRelationDirection::Subtype => {

ecx.sub(param_env, alias_lhs, alias_rhs)?;

}

}

ecx.evaluate_added_goals_and_make_canonical_response(Certainty::Yes)

})

}

Since we can't just use the generic eq and sub methods on EvalCtxt.

compiler-errors · 2023-09-28T22:55:49Z

I'm somewhat skeptical of the perf regression, but maybe it's real? Diesel looks spiky recently.

compiler-errors · 2023-10-04T04:22:50Z

Rebased and testing again

@bors try @rust-timer queue

bors · 2023-10-04T04:23:59Z

⌛ Trying commit be29d22 with merge 0027216...

…-variance, r=<try> Relate alias ty with variance In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀. This change is a bit more invasive, but I'd rather not special-case it [here](https://github.com/rust-lang/rust/blob/aeaa5c30e5c9041264a2e8314b68ad84c2dc3169/compiler/rustc_trait_selection/src/solve/alias_relate.rs#L171-L190) and then have it break elsewhere. I'm doing a perf run to see if the extra call to `def_kind` is that expensive, if it is, I'll reconsider. r? `@lcnr`

bors · 2023-10-04T05:38:33Z

☀️ Try build successful - checks-actions
Build commit: 0027216 (0027216f8bc36b3488f0390f02948671465ff957)

rust-timer · 2023-10-04T08:10:03Z

Finished benchmarking commit (0027216): comparison URL.

Overall result: ✅ improvements - 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 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
Improvements ✅ (secondary)	-0.9%	[-1.1%, -0.7%]	7
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

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

Cycles

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)	1.4%	[1.4%, 1.4%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.2%	[-3.2%, -3.2%]	1
All ❌✅ (primary)	1.4%	[1.4%, 1.4%]	1

Binary size

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

Bootstrap: 622.307s -> 623.821s (0.24%)
Artifact size: 272.04 MiB -> 272.04 MiB (0.00%)

lcnr · 2023-10-11T13:00:39Z

@bors r+ rollup (new solver)

bors · 2023-10-11T13:00:41Z

📌 Commit be29d22 has been approved by lcnr

It is now in the queue for this repository.

…th-variance, r=lcnr Relate alias ty with variance In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀. This change is a bit more invasive, but I'd rather not special-case it [here](https://github.com/rust-lang/rust/blob/aeaa5c30e5c9041264a2e8314b68ad84c2dc3169/compiler/rustc_trait_selection/src/solve/alias_relate.rs#L171-L190) and then have it break elsewhere. I'm doing a perf run to see if the extra call to `def_kind` is that expensive, if it is, I'll reconsider. r? `@lcnr`

…iaskrgr Rollup of 5 pull requests Successful merges: - rust-lang#116219 (Relate alias ty with variance) - rust-lang#116315 (Do not check for impossible predicates in const-prop lint.) - rust-lang#116436 (Structurally normalize for closure) - rust-lang#116597 (Prevent showing methods from blanket impls of not available foreign traits to show up in the search results) - rust-lang#116627 (small cleanup) r? `@ghost` `@rustbot` modify labels: rollup

Rollup merge of rust-lang#116219 - compiler-errors:relate-alias-ty-with-variance, r=lcnr Relate alias ty with variance In the new solver, turns out that the subst-relate branch of the alias-relate predicate was relating args invariantly even for opaques, which have variance 💀. This change is a bit more invasive, but I'd rather not special-case it [here](https://github.com/rust-lang/rust/blob/aeaa5c30e5c9041264a2e8314b68ad84c2dc3169/compiler/rustc_trait_selection/src/solve/alias_relate.rs#L171-L190) and then have it break elsewhere. I'm doing a perf run to see if the extra call to `def_kind` is that expensive, if it is, I'll reconsider. r? ``@lcnr``

rustbot assigned lcnr Sep 28, 2023

This comment has been minimized.

Sign in to view

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. S-waiting-on-perf Status: Waiting on a perf run to be completed. labels Sep 28, 2023

rust-cloud-vms bot force-pushed the relate-alias-ty-with-variance branch from e32b697 to f310e3c Compare September 28, 2023 06:44

This comment was marked as duplicate.

Sign in to view

This comment was marked as outdated.

Sign in to view

fmease reviewed Sep 28, 2023

View reviewed changes

This comment has been minimized.

Sign in to view

rustbot added perf-regression Performance regression. and removed S-waiting-on-perf Status: Waiting on a perf run to be completed. labels Sep 28, 2023

lcnr reviewed Sep 28, 2023

View reviewed changes

compiler-errors added 3 commits October 4, 2023 04:22

Relate AliasTy considering variance

5087bb1

Remove unnecessary relate impl

cdef3b1

Make it clear that args default to being related invariantly

be29d22

rust-cloud-vms bot force-pushed the relate-alias-ty-with-variance branch from f310e3c to be29d22 Compare October 4, 2023 04:22

This comment has been minimized.

Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Oct 4, 2023

This comment has been minimized.

Sign in to view

rustbot removed S-waiting-on-perf Status: Waiting on a perf run to be completed. perf-regression Performance regression. labels Oct 4, 2023

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Oct 11, 2023

This was referenced Oct 11, 2023

Rollup of 5 pull requests #116636

Closed

Rollup of 5 pull requests #116640

Merged

bors merged commit 8ddc0df into rust-lang:master Oct 11, 2023
12 checks passed

rustbot added this to the 1.75.0 milestone Oct 11, 2023

Dajamante mentioned this pull request Oct 12, 2023

Pull upstream master 2023 10 12 ferrocene/ferrocene#48

Merged

matthiaskrgr mentioned this pull request Oct 13, 2023

ice: unknown alias DefKind: AnonConst #116710

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Relate alias ty with variance #116219

Relate alias ty with variance #116219

compiler-errors commented Sep 28, 2023 •

edited

Loading

compiler-errors commented Sep 28, 2023

This comment has been minimized.

bors commented Sep 28, 2023

This comment was marked as duplicate.

compiler-errors commented Sep 28, 2023

bors commented Sep 28, 2023

bors commented Sep 28, 2023

This comment was marked as outdated.

fmease Sep 28, 2023

compiler-errors Sep 28, 2023

bors commented Sep 28, 2023

This comment has been minimized.

rust-timer commented Sep 28, 2023

lcnr Sep 28, 2023

compiler-errors Sep 28, 2023

lcnr Sep 29, 2023

compiler-errors Sep 29, 2023

compiler-errors commented Sep 28, 2023

compiler-errors commented Oct 4, 2023

This comment has been minimized.

bors commented Oct 4, 2023

bors commented Oct 4, 2023

This comment has been minimized.

rust-timer commented Oct 4, 2023

lcnr commented Oct 11, 2023

bors commented Oct 11, 2023

	fn assemble_subst_relate_candidate(
	&mut self,
	param_env: ty::ParamEnv<'tcx>,
	alias_lhs: ty::AliasTy<'tcx>,
	alias_rhs: ty::AliasTy<'tcx>,
	direction: ty::AliasRelationDirection,
	) -> QueryResult<'tcx> {
	self.probe_misc_candidate("args relate").enter(\|ecx\| {
	match direction {
	ty::AliasRelationDirection::Equate => {
	ecx.eq(param_env, alias_lhs, alias_rhs)?;
	}
	ty::AliasRelationDirection::Subtype => {
	ecx.sub(param_env, alias_lhs, alias_rhs)?;
	}
	}

	ecx.evaluate_added_goals_and_make_canonical_response(Certainty::Yes)
	})
	}

Relate alias ty with variance #116219

Relate alias ty with variance #116219

Conversation

compiler-errors commented Sep 28, 2023 • edited Loading

compiler-errors commented Sep 28, 2023

This comment has been minimized.

bors commented Sep 28, 2023

This comment was marked as duplicate.

compiler-errors commented Sep 28, 2023

bors commented Sep 28, 2023

bors commented Sep 28, 2023

This comment was marked as outdated.

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bors commented Sep 28, 2023

This comment has been minimized.

rust-timer commented Sep 28, 2023

Overall result: ❌✅ regressions and improvements - ACTION NEEDED

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

compiler-errors commented Sep 28, 2023

compiler-errors commented Oct 4, 2023

This comment has been minimized.

bors commented Oct 4, 2023

bors commented Oct 4, 2023

This comment has been minimized.

rust-timer commented Oct 4, 2023

Overall result: ✅ improvements - no action needed

lcnr commented Oct 11, 2023

bors commented Oct 11, 2023

compiler-errors commented Sep 28, 2023 •

edited

Loading