`-Znext-solver` filter out some uninteresting opaques while canonicalization by ShoyuVanilla · Pull Request #164368 · rust-lang/rust

ShoyuVanilla · 2026-10-15T15:30:19Z

Context: #t-types/trait-system-refactor > higher ranked opaque type perf

Partially extracted from #161414

rust-lang/trait-system-refactor-initiative#242 is not fixed by this. The overflow goes away by this, but as I wrote in the above zulip stream, hr opaque types aren't replaced by infer vars and that prevents the pending obligation containing it from being fully normalized and solved in the root context.

r? adwinwhite as we've talked on this in the zulip stream (cc @lcnr I guess you might be busy in EuroRust 😅)

rustbot · 2026-10-15T15:30:26Z

adwinwhite is currently at their maximum review capacity.
They may take a while to respond.

ShoyuVanilla · 2026-10-15T15:31:06Z


 fn build_multiple<'a>() -> impl Sized {
    spawn(async { build_dependencies().await });
-    //[next]~^ ERROR overflow evaluating the requirement


This isn't overflowing anymore and I guess this is an unexpected improvement

View changes since the review

Actually, this was in the triage list as a regression 😅
https://hackmd.io/RQF9zqx8R3aBl_-gr2YU6w?view#Pass--gt-Regular-Fail

ShoyuVanilla · 2026-10-15T15:31:23Z

@bors try @rust-timer queue

`-Znext-solver` filter out some uninteresting opaques while canonicalization

…ization

rust-bors · 2026-10-15T16:34:50Z

☀️ Try build successful (CI)
Build commit: 307e94e (307e94edc04512006422385a8ce280bca6e7ed55)
Base parent: 3b13d3f (3b13d3fd58d21a6ffd3aa6dec7ae6b299405acd1)

rust-timer · 2026-10-15T17:54:51Z

Finished benchmarking commit (307e94e): comparison URL.

Overall result: ❌✅ regressions and improvements - please read:

Benchmarking means the PR may be perf-sensitive. Consider adding rollup=never if this change is not fit for rolling up.

Next, please: If you can, justify the regressions found in this try perf run in writing along with @rustbot label: +perf-regression-triaged. If not, fix the regressions and do another perf run. Neutral or positive results will clear the label automatically.

@bors rollup=iffy rustc-perf
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.3%	[0.2%, 0.4%]	6
Improvements ✅ (primary)	-0.8%	[-0.8%, -0.8%]	1
Improvements ✅ (secondary)	-1.7%	[-3.5%, -0.6%]	7
All ❌✅ (primary)	-0.8%	[-0.8%, -0.8%]	1

Max RSS (memory usage)

Results (primary 1.2%, secondary 3.5%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	1.2%	[1.2%, 1.2%]	1
Regressions ❌ (secondary)	3.5%	[3.1%, 4.0%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.2%	[1.2%, 1.2%]	1

Cycles

Results (primary -2.5%, secondary 1.6%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	7.3%	[6.7%, 8.5%]	3
Improvements ✅ (primary)	-2.5%	[-2.5%, -2.5%]	1
Improvements ✅ (secondary)	-2.7%	[-3.4%, -2.0%]	4
All ❌✅ (primary)	-2.5%	[-2.5%, -2.5%]	1

Binary size

Results (primary 0.0%, secondary 0.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	0.0%	[0.0%, 0.0%]	1
Regressions ❌ (secondary)	0.0%	[0.0%, 0.0%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.0%	[0.0%, 0.0%]	1

Bootstrap: 492.307s -> 489.052s (-0.66%)
Artifact size: 407.14 MiB -> 409.22 MiB (0.51%)

adwinwhite · 2026-10-16T10:23:16Z

Does this help with other compile-time regressions in the ui test triage table?

ShoyuVanilla · 2026-10-16T10:25:49Z

Let me check (completely forgot about it while pondering upon hr opaques 😅 )

ShoyuVanilla · 2026-10-16T17:00:50Z

Yeah, all four tests in compiletime and one test in hang for it compiles instataneously

adwinwhite

Sorry, not much actionable feedback. I guess this needs some discussions to make the impl easier to reason.

View changes since this review

adwinwhite · 2026-10-20T07:26:25Z

+    /// Filter out not-so-useful opaque types, i.e. whose hidden type is not
+    /// mentioned in `variables` so far and has any var within opaque type key
+    /// that is not mentioned in the same manner. This helps the perf by avoiding
+    /// having many inputs/outputs differing only on some not interesting opaques
+    /// and even saves us from overflowing due to having too many opaques.


What's the reasoning of defining not-so-useful like this? 🤔

I thought we could drop opaques whose hidden types are infers and not in var_values. And we'd also keep opaques whose hidden infers are mentioned in other opaque keys for the reason you said above the fixpoint computation. The latter part feels slightly hacky though.

adwinwhite · 2026-10-20T07:30:39Z

-            } else {
-                predefined_opaques_in_body
-            };
+            rest_canonicalizer.filter_and_canonicalize_opaque_types(predefined_opaques_in_body);


filtering in response should be enough to prevent overflows?
uncertain about whether this filters out "useful` opaques. I need to think more on this.

Yeah, that would be enough for preventing overflows and from the input side it would help only the perf due to better cache hits.
On what is useful and what is useless opaque (for both this and your above comment), yeah, I have sth in my mind but haven't tried to clearly state them anywhere. I'll try that 😅

adwinwhite · 2026-10-20T07:33:24Z

+            }
+        }
+
+        // This should be synced with `Canonicalizer`'s `TypeFolder` implementation.


We also have a mentioned_region visitor in filter_irrelevant_region_constraints.
It feels disorganized to have them in multiple places.

adwinwhite · 2026-10-20T07:37:30Z

Yeah, all four tests in compiletime and one test in hang for it compiles instataneously

We could add next solver revisions for them then :>

rustbot assigned adwinwhite Oct 15, 2026

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. WG-trait-system-refactor The Rustc Trait System Refactor Initiative (-Znext-solver) labels Oct 15, 2026

ShoyuVanilla commented Oct 15, 2026

View reviewed changes

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 15, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Oct 15, 2026

Auto merge of #164368 - ShoyuVanilla:opaca-parva, r=<try>

307e94e

`-Znext-solver` filter out some uninteresting opaques while canonicalization

rust-cloud-vms Bot force-pushed the opaca-parva branch from c7c7a1a to 7ff1473 Compare October 15, 2026 15:43

-Znext-solver filter out some uninteresting opaques while canonical…

eea7880

…ization

rust-cloud-vms Bot force-pushed the opaca-parva branch from 7ff1473 to eea7880 Compare October 15, 2026 15:45

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 Oct 15, 2026

adwinwhite reviewed Oct 20, 2026

View reviewed changes

Conversation

ShoyuVanilla commented Oct 15, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Oct 15, 2026

Uh oh!

ShoyuVanilla Oct 15, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ShoyuVanilla commented Oct 15, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

rust-bors Bot commented Oct 15, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Oct 15, 2026

Overall result: ❌✅ regressions and improvements - please read:

Uh oh!

adwinwhite commented Oct 16, 2026

Uh oh!

ShoyuVanilla commented Oct 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ShoyuVanilla commented Oct 16, 2026

Uh oh!

adwinwhite left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

adwinwhite commented Oct 20, 2026

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

ShoyuVanilla commented Oct 15, 2026 •

edited by rustbot

Loading

ShoyuVanilla Oct 15, 2026 •

edited by rustbot

Loading

ShoyuVanilla commented Oct 16, 2026 •

edited

Loading

adwinwhite left a comment •

edited by rustbot

Loading