Uplift next trait solver to `rustc_next_trait_solver` #126614

compiler-errors · 2024-06-18T01:04:18Z

🎉

There's so many FIXMEs! Sorry! Ideally this merges with the FIXMEs and we track and squash them over the near future.

Also, this still doesn't build on anything other than rustc. I still need to fix feature = "nightly" in rustc_type_ir, and remove and fix all the nightly feature usage in the new trait solver (notably: let-chains).

Also, sorry @lcnr I know you asked for me to separate the commit where we mv rustc_trait_selection/solve/... rustc_next_trait_solver/solve/..., but I had already done all the work by that point. Luckily, git understands the file moves so it should still be relatively reviewable.

If this is still very difficult to review, then I can do some rebasing magic to try to separate this out. Please let me know!

r? lcnr

rustbot · 2024-06-18T01:04:27Z

Some changes occurred to the core trait solver

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

These commits modify the Cargo.lock file. Unintentional changes to Cargo.lock can be introduced when switching branches and rebasing PRs.

If this was unintentional then you should revert the changes before this PR is merged.
Otherwise, you can ignore this comment.

compiler-errors · 2024-06-18T01:07:09Z

compiler/rustc_infer/src/infer/mod.rs

@@ -334,149 +334,6 @@ pub struct InferCtxt<'tcx> {
    pub obligation_inspector: Cell<Option<ObligationInspector<'tcx>>>,
 }

-impl<'tcx> ty::InferCtxtLike for InferCtxt<'tcx> {


This was moved to rustc_trait_selection/src/solve/infcx.rs

compiler-errors · 2024-06-18T01:07:52Z

compiler/rustc_infer/src/infer/mod.rs

@@ -830,6 +687,10 @@ impl<'tcx> InferCtxt<'tcx> {
        self.tcx.dcx()
    }

+    pub fn defining_opaque_types(&self) -> &'tcx ty::List<LocalDefId> {


Since defining_opaque_types is a private field on InferCtxt, and I've chosen to expose only one trait (SolverDelegate) and not two (InferCtxtLike), we gotta make it possible to reference this in the impl in rustc_trait_solver.

compiler-errors · 2024-06-18T01:09:33Z

compiler/rustc_middle/src/traits/solve/cache.rs

@@ -48,7 +42,7 @@ impl<'tcx> EvaluationCache<'tcx> {
        if cfg!(debug_assertions) {
            drop(map);
            let expected = CacheData { result, proof_tree, additional_depth, encountered_overflow };
-            let actual = self.get(tcx, key, [], Limit(additional_depth));


Limit is defined in rustc_session, and I was kinda too lazy to find a good ancestor crate to put Limit in...

Could fix it as a follow-up or fix it in this PR if necessary.

compiler-errors · 2024-06-18T01:10:24Z

compiler/rustc_middle/src/ty/context.rs

    type GenericArgs = ty::GenericArgsRef<'tcx>;
+


Sorry, whenever I use rust-analyzer to sort items by definition, the spacing gets all fucked up.

compiler-errors · 2024-06-18T01:14:03Z

compiler/rustc_next_trait_solver/src/solve/assembly/mod.rs

@@ -301,16 +302,14 @@ impl<'tcx> EvalCtxt<'_, InferCtxt<'tcx>> {

        let mut candidates = vec![];

-        self.assemble_non_blanket_impl_candidates(goal, &mut candidates);
+        self.assemble_impl_candidates(goal, &mut candidates);


This consolidates assemble_{blanket,non_blanket}_impl_candidates. See the definition of for_each_relevant_impl that I added to Interner.

compiler-errors · 2024-06-18T01:29:23Z

compiler/rustc_next_trait_solver/src/solve/search_graph.rs

    /// When we encounter a coinductive cycle, we have to fetch the
    /// result of that cycle while we are still computing it. Because
    /// of this we continuously recompute the cycle until the result
    /// of the previous iteration is equal to the final result, at which
    /// point we are done.
-    fn fixpoint_step_in_task<F>(
+    fn fixpoint_step_in_task<Infcx, F>(


Could make the search graph generic over Infcx and use a PhantomData just like we do with ProofTreeBuilder?

compiler-errors · 2024-06-18T01:30:11Z

compiler/rustc_next_trait_solver/src/solve/search_graph.rs

+};
+
+#[derive(Copy, Clone, PartialEq, Eq, Debug)]
+pub struct SolverLimit(usize);


Ad-hoc reintroduced a Limit-like newtype. We should probably consolidate this with the normal Limit. Probably best left for a follow-up.

compiler/rustc_next_trait_solver/src/solve/normalizes_to/opaque_types.rs

compiler-errors · 2024-06-18T01:31:40Z

compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

-        let eligible = if node_item.is_final() {
-            // Non-specializable items are always projectable.
-            true
+        Ok(if target_container_def_id == impl_trait_ref.def_id {


This logic was rewritten to avoid LeafDef and Node.

compiler-errors · 2024-06-18T01:33:11Z

compiler/rustc_next_trait_solver/src/solve/normalizes_to/mod.rs

        let tcx = ecx.interner();

        let goal_trait_ref = goal.predicate.alias.trait_ref(tcx);
-        let impl_trait_header = tcx.impl_trait_header(impl_def_id).unwrap();


I split impl_trait_header back into impl_trait_ref and impl_polarity b/c I couldn't be bothered to uplift ImplHeader. I can leave a FIXME to add it back if it matters that much.

lcnr

nice nice

some nits, then r=me

good work

@bors rollup=never

compiler/rustc_middle/src/ty/context.rs

compiler/rustc_middle/src/ty/generic_args.rs

compiler/rustc_next_trait_solver/src/solve/assembly/mod.rs

compiler/rustc_next_trait_solver/src/solve/eval_ctxt/canonical.rs

compiler/rustc_next_trait_solver/src/solve/normalizes_to/opaque_types.rs

compiler/rustc_next_trait_solver/src/solve/trait_goals.rs

lcnr · 2024-06-18T13:50:28Z

compiler/rustc_trait_selection/src/solve/fulfill.rs

@@ -288,7 +293,10 @@ fn fulfillment_error_for_stalled<'tcx>(
    root_obligation: PredicateObligation<'tcx>,
 ) -> FulfillmentError<'tcx> {
    let (code, refine_obligation) = infcx.probe(|_| {
-        match infcx.evaluate_root_goal(root_obligation.clone().into(), GenerateProofTree::No).0 {
+        match <&SolverDelegate<'tcx>>::from(infcx)
+            .evaluate_root_goal(root_obligation.clone().into(), GenerateProofTree::No)


This seems very meh, I would prefer to implement evaluate_root_goal and evaluate_root_goal_raw via an extension trait on InferCtxt. Why do we need the SolverDelegate super trait on that one?

because EvalCtxt::enter_root requires a SolverDelegate

but enter_root is only used inside of evaluate_root_goal, so can't we wrap the InferCtxt inside of that function instead?

compiler/rustc_type_ir/src/lang_items.rs

compiler-errors · 2024-06-18T15:23:50Z

@bors r=lcnr rollup=never p=1 (ever so slightly bitrotty)

bors · 2024-06-18T15:23:54Z

📌 Commit 6609501 has been approved by lcnr

It is now in the queue for this repository.

bors · 2024-06-18T19:41:36Z

⌛ Testing commit 6609501 with merge 8fcd4dd...

bors · 2024-06-18T22:03:19Z

☀️ Test successful - checks-actions
Approved by: lcnr
Pushing 8fcd4dd to master...

rust-timer · 2024-06-19T01:14:20Z

Finished benchmarking commit (8fcd4dd): 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 (primary 0.8%)

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)	3.8%	[3.8%, 3.8%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.1%	[-2.1%, -2.1%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.8%	[-2.1%, 3.8%]	2

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: 671.341s -> 691.815s (3.05%)
Artifact size: 320.42 MiB -> 323.69 MiB (1.02%)

rustbot assigned lcnr Jun 18, 2024

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 Jun 18, 2024

This comment has been minimized.

Sign in to view

compiler-errors commented Jun 18, 2024

View reviewed changes

compiler-errors force-pushed the uplift-next-trait-solver branch from 06e3d9b to 708a54d Compare June 18, 2024 02:00

lcnr reviewed Jun 18, 2024

View reviewed changes

compiler-errors added 4 commits June 18, 2024 10:40

Make SearchGraph fully generic

dba4147

Uplift PredefinedOpaquesData

e7d2d95

SolverDelegate

baf94bd

Uplift the new trait solver

532149e

compiler-errors force-pushed the uplift-next-trait-solver branch from 708a54d to e3e76a5 Compare June 18, 2024 14:55

compiler-errors added 3 commits June 18, 2024 11:04

Fix impl for SolverDelegate

7d2be88

Explicitly import tracing macros

fb6f4b4

Fix transmute goal

6609501

compiler-errors force-pushed the uplift-next-trait-solver branch from e3e76a5 to 6609501 Compare June 18, 2024 15:04

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 Jun 18, 2024

bors added the merged-by-bors This PR was explicitly merged by bors. label Jun 18, 2024

bors merged commit 8fcd4dd into rust-lang:master Jun 18, 2024
7 checks passed

rustbot added this to the 1.81.0 milestone Jun 18, 2024

This was referenced Jun 18, 2024

alias-relate: add fast reject optimization #124852

Open

do not rerun trait solver cycles on ambiguity #125981

Closed

lcnr mentioned this pull request Sep 4, 2024

Tracking Issue for -Znext-solver #107374

Open

14 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uplift next trait solver to `rustc_next_trait_solver` #126614

Uplift next trait solver to `rustc_next_trait_solver` #126614

compiler-errors commented Jun 18, 2024 •

edited

Loading

rustbot commented Jun 18, 2024

This comment has been minimized.

compiler-errors Jun 18, 2024

compiler-errors Jun 18, 2024

compiler-errors Jun 18, 2024

compiler-errors Jun 18, 2024

compiler-errors Jun 18, 2024

compiler-errors Jun 18, 2024

compiler-errors Jun 18, 2024 •

edited

Loading

compiler-errors Jun 18, 2024

compiler-errors Jun 18, 2024

lcnr left a comment

lcnr Jun 18, 2024

compiler-errors Jun 18, 2024

lcnr Jun 18, 2024

compiler-errors commented Jun 18, 2024

bors commented Jun 18, 2024

bors commented Jun 18, 2024

bors commented Jun 18, 2024

rust-timer commented Jun 19, 2024

Uplift next trait solver to rustc_next_trait_solver #126614

Uplift next trait solver to rustc_next_trait_solver #126614

Conversation

compiler-errors commented Jun 18, 2024 • edited Loading

rustbot commented Jun 18, 2024

This comment has been minimized.

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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

compiler-errors Jun 18, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lcnr left a comment

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 Jun 18, 2024

bors commented Jun 18, 2024

bors commented Jun 18, 2024

bors commented Jun 18, 2024

rust-timer commented Jun 19, 2024

Overall result: no relevant changes - no action needed

Uplift next trait solver to `rustc_next_trait_solver` #126614

Uplift next trait solver to `rustc_next_trait_solver` #126614

compiler-errors commented Jun 18, 2024 •

edited

Loading

compiler-errors Jun 18, 2024 •

edited

Loading