rustc: allow non-empty ParamEnv's in global trait select/eval caches. by eddyb · Pull Request #66821 · rust-lang/rust

eddyb · 2019-11-27T19:15:43Z

Based on #66963

This appears to alleviate the symptoms of #65510 locally (without fixing WF directly), and is potentially easier to validate as sound (since it's a more ad-hoc version of queries we already have).

I'm opening this PR primarily to test the effects on perf.

r? @nikomatsakis cc @rust-lang/wg-traits

eddyb · 2019-11-27T19:16:11Z

@bors try @rust-timer queue

rust-timer · 2019-11-27T19:16:13Z

Awaiting bors try build completion

bors · 2019-11-27T19:16:22Z

⌛ Trying commit d637c85 with merge 06db465248f3aa76f16924e778609dec610063a4...

bors · 2019-11-27T22:04:07Z

☀️ Try build successful - checks-azure
Build commit: 06db465248f3aa76f16924e778609dec610063a4 (06db465248f3aa76f16924e778609dec610063a4)

rust-timer · 2019-11-27T22:04:09Z

Queued 06db465248f3aa76f16924e778609dec610063a4 with parent 04e69e4, future comparison URL.

rust-timer · 2019-11-28T01:17:37Z

Finished benchmarking try commit 06db465248f3aa76f16924e778609dec610063a4, comparison URL.

eddyb · 2019-11-28T13:10:32Z

Looks like the wins aren't as significant as #66020 (comment), but they are comparable.

@nikomatsakis

[WIP] [DO NOT MERGE] combine #66020 and #66821. That is, the two fixes for #65510, and only for perf testing purposes. The fact that they both work to a comparable extent, while touching different parts of the trait system, made me curious if there would be any gains from having both. r? @nikomatsakis

eddyb · 2019-11-28T22:28:21Z

Oh, also, the max-rss for the above run has increased across the board (because we're caching more), but it's not as bad as I was worried it might be.

Then again, looking at #66020's own max-rss, I'd argue it's noisy, because that PR only removes work, so I don't see why it could/would increase memory usage.

eddyb · 2019-12-02T17:12:27Z

Take 2 (after getting tests to pass):
@bors try @rust-timer queue

rust-timer · 2019-12-02T17:12:29Z

Awaiting bors try build completion

bors · 2019-12-02T17:12:38Z

⌛ Trying commit a266ea0 with merge 1ff0441...

@nikomatsakis

rustc: allow non-empty ParamEnv's in global trait select/eval caches. *Based on #66963* This appears to alleviate the symptoms of #65510 locally (without fixing WF directly), and is potentially easier to validate as sound (since it's a more ad-hoc version of queries we already have). I'm opening this PR primarily to test the effects on perf. r? @nikomatsakis cc @rust-lang/wg-traits

bors · 2019-12-02T19:54:11Z

☀️ Try build successful - checks-azure
Build commit: 1ff0441 (1ff04410af642dde1480ae29b085544c2d05c33c)

rust-timer · 2019-12-02T19:54:13Z

Queued 1ff0441 with parent 2da942f, future comparison URL.

rust-timer · 2019-12-03T02:43:43Z

Finished benchmarking try commit 1ff0441, comparison URL.

eddyb · 2019-12-03T09:49:36Z

Looks comparable, so the changes I had to make didn't cause much of a difference.

Dylan-DPC-zz · 2019-12-10T13:22:39Z

Blocked on #66963

nikomatsakis · 2019-12-10T23:03:30Z

So I left a comment on #66963 explaining that I have some mild trepidation about the behavior of projection when inference variables are involved, but I'm inclined to land this PR since I don't think it will do anything incorrect.

@bors r+

bors · 2019-12-10T23:03:32Z

📌 Commit a266ea0 has been approved by nikomatsakis

eddyb · 2019-12-10T23:13:57Z

@bors rollup=never

bors · 2019-12-11T12:29:46Z

⌛ Testing commit a266ea0 with merge 90b957a...

@nikomatsakis

rustc: allow non-empty ParamEnv's in global trait select/eval caches. *Based on #66963* This appears to alleviate the symptoms of #65510 locally (without fixing WF directly), and is potentially easier to validate as sound (since it's a more ad-hoc version of queries we already have). I'm opening this PR primarily to test the effects on perf. r? @nikomatsakis cc @rust-lang/wg-traits

RalfJung · 2019-12-11T12:34:01Z

+LL | fn foo_desugared<T: TraitWithAssoc>(_: T) -> Foo<T::Assoc> {
+ | -- help: consider further restricting this bound: `T: TraitWithAssoc +`


This error looks pretty wrong though, doesn't it?

Yeah, but it makes sense in the context.
It's the error you get if the other errors don't prevent the impl Trait from getting the incompatible concrete type, because of the type parameter that is now in the wrong ParamEnv, which doesn't have the original bound.
This is pre-existing, it was just hidden by the bug #66963 fixes.

bors · 2019-12-11T15:44:13Z

☀️ Test successful - checks-azure
Approved by: nikomatsakis
Pushing 90b957a to master...

rust-highfive assigned nikomatsakis Nov 27, 2019

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Nov 27, 2019

eddyb mentioned this pull request Nov 27, 2019
Quadratic slowdown for lookup of bounds on associated types. #65510
Open

This comment has been minimized.
Sign in to view

eddyb mentioned this pull request Nov 28, 2019
[WIP] [DO NOT MERGE] combine #66020 and #66821. #66838
Closed

eddyb force-pushed the global-trait-caching branch 2 times, most recently from 8f1af1d to 36b23edCompare December 2, 2019 15:44

eddyb mentioned this pull request Dec 2, 2019
rustc: include ParamEnv in global trait select/eval cache keys. #66963
Merged

rustc: include ParamEnv in global trait select/eval cache keys.
1d1298e

rustc: allow non-empty ParamEnv's in global trait select/eval caches.
a266ea0

eddyb force-pushed the global-trait-caching branch from 36b23ed to a266ea0Compare December 2, 2019 16:54

eddyb changed the title ~~[WIP] rustc: experiment with ParamEnv-aware caching in traits::select.~~rustc: allow non-empty ParamEnv's in global trait select/eval caches.Dec 2, 2019

Dylan-DPC-zz added S-blocked Status: Blocked on something else such as an RFC or other implementation work. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Dec 10, 2019

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-blocked Status: Blocked on something else such as an RFC or other implementation work. labels Dec 10, 2019

RalfJung reviewed Dec 11, 2019
View reviewed changes

bors added the merged-by-bors This PR was explicitly merged by bors. label Dec 11, 2019

bors merged commit a266ea0 into rust-lang:masterDec 11, 2019

This was referenced Dec 11, 2019
Implement RFC 2532 – Associated Type Defaults #61812
Merged
Allocate HIR on an arena 4/4 #67032
Merged

eddyb deleted the global-trait-caching branch December 11, 2019 17:10

eddyb mentioned this pull request Dec 12, 2019
[WIP] rustc: avoid checking Trait's predicates for WF(<T as Trait>::X). #66020
Closed

Aaron1011 mentioned this pull request Jan 17, 2020
ICE in librustc::traits::select::confirm_param_candidate #68295
Closed

matthiaskrgr mentioned this pull request Jul 8, 2026
[ICE]: monomorphization failure #158756
Open

		LL \| fn foo_desugared<T: TraitWithAssoc>(_: T) -> Foo<T::Assoc> {
		\| -- help: consider further restricting this bound: `T: TraitWithAssoc +`

Uh oh!

Uh oh!

Conversation

eddyb commented Nov 27, 2019• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

eddyb commented Nov 27, 2019

Uh oh!

rust-timer commented Nov 27, 2019

Uh oh!

bors commented Nov 27, 2019

Uh oh!

This comment has been minimized.

bors commented Nov 27, 2019

Uh oh!

rust-timer commented Nov 27, 2019

Uh oh!

rust-timer commented Nov 28, 2019

Uh oh!

eddyb commented Nov 28, 2019

Uh oh!

eddyb commented Nov 28, 2019

Uh oh!

eddyb commented Dec 2, 2019

Uh oh!

rust-timer commented Dec 2, 2019

Uh oh!

bors commented Dec 2, 2019

Uh oh!

bors commented Dec 2, 2019

Uh oh!

rust-timer commented Dec 2, 2019

Uh oh!

rust-timer commented Dec 3, 2019

Uh oh!

eddyb commented Dec 3, 2019

Uh oh!

Dylan-DPC-zz commented Dec 10, 2019

Uh oh!

nikomatsakis commented Dec 10, 2019

Uh oh!

bors commented Dec 10, 2019

Uh oh!

eddyb commented Dec 10, 2019

Uh oh!

bors commented Dec 11, 2019

Uh oh!

RalfJungDec 11, 2019

Choose a reason for hiding this comment

Uh oh!

eddybDec 11, 2019

Choose a reason for hiding this comment

Uh oh!

bors commented Dec 11, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

eddyb commented Nov 27, 2019•
edited
Loading