name resolution for guard patterns #140746

dianne · 2025-05-07T14:37:39Z

This PR provides an initial implementation of name resolution for guard patterns (RFC 3637). This does not change the requirement that the bindings on either side of an or-pattern must be the same (proposal here); the code that handles that is separate from what this PR touches, so I'm saving it for a follow-up.

On a technical level, this separates "collecting the bindings in a pattern" (which was already done for or-patterns) from "introducing those bindings into scope". I believe the approach used here can be extended straightforwardly in the future to work with if let guard patterns, but I haven't tried it myself since we don't allow those yet.

Tracking issue for guard patterns: #129967

cc @Nadrieril

I'll be modifying it in future commits, so I think it's cleanest to abstract it out. Possibly a newtype would be ideal, but for now this is least disruptive.

This splits introduction of bindings into scope (`apply_pattern_bindings`) apart from manipulation of the pattern's binding map (`fresh_binding`). By delaying the latter, we can keep bindings from appearing in-scope in guards. Since `fresh_binding` is now specifically for manipulating a pattern's bindings map, this commit also inlines a use of `fresh_binding` that was only adding to the innermost rib.

rustbot · 2025-05-07T14:37:45Z

r? @oli-obk

rustbot has assigned @oli-obk.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Nadrieril · 2025-05-08T09:28:57Z

@bors try @rust-timer queue

name resolution for guard patterns This PR provides an initial implementation of name resolution for guard patterns [(RFC 3637)](https://github.com/rust-lang/rfcs/blob/master/text/3637-guard-patterns.md). This does not change the requirement that the bindings on either side of an or-pattern must be the same [(proposal here)](https://github.com/rust-lang/rfcs/blob/master/text/3637-guard-patterns.md#allowing-mismatching-bindings-when-possible); the code that handles that is separate from what this PR touches, so I'm saving it for a follow-up. On a technical level, this separates "collecting the bindings in a pattern" (which was already done for or-patterns) from "introducing those bindings into scope". I believe the approach used here can be extended straightforwardly in the future to work with `if let` guard patterns, but I haven't tried it myself since we don't allow those yet. Tracking issue for guard patterns: rust-lang#129967 cc `@Nadrieril`

bors · 2025-05-08T09:30:09Z

⌛ Trying commit d7ac436 with merge f04862f...

bors · 2025-05-08T11:37:16Z

☀️ Try build successful - checks-actions
Build commit: f04862f (f04862fbe36c5246e1b11151b86d39c1ae766cb5)

rust-timer · 2025-05-08T13:15:03Z

Finished benchmarking commit (f04862f): comparison URL.

Overall result: ❌ regressions - please read the text below

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 the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	1.0%	[1.0%, 1.0%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.0%	[1.0%, 1.0%]	1

Max RSS (memory usage)

Results (primary 0.5%, secondary -1.3%)

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)	0.7%	[0.5%, 1.4%]	20
Regressions ❌ (secondary)	1.6%	[1.6%, 1.6%]	1
Improvements ✅ (primary)	-1.3%	[-1.8%, -0.9%]	2
Improvements ✅ (secondary)	-2.8%	[-3.4%, -2.2%]	2
All ❌✅ (primary)	0.5%	[-1.8%, 1.4%]	22

Cycles

Results (primary -0.6%, secondary -1.8%)

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)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.6%	[-1.0%, -0.4%]	30
Improvements ✅ (secondary)	-1.8%	[-1.8%, -1.8%]	1
All ❌✅ (primary)	-0.6%	[-1.0%, -0.4%]	30

Binary size

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

Bootstrap: 769.201s -> 769.203s (0.00%)
Artifact size: 365.25 MiB -> 365.08 MiB (-0.05%)

dianne · 2025-05-08T13:57:39Z

I might know how to get perf back to where it was before and trim down this PR's diff. I'll try pushing an update later. @rustbot author

rustbot · 2025-05-08T13:57:43Z

Reminder, once the PR becomes ready for a review, use @rustbot ready.

dianne · 2025-05-09T06:21:47Z

Actually, I don't think I'm going to try improving the perf yet. I have no idea what's causing the instruction count regression in nalgebra. That's the only instruction count regression, and I don't see how this change could have caused it. late_resolve_crate is only 1.46% of the total execution already, so getting 1% more instructions from this change would be pretty wild, especially considering the time delta in the perf report for late_resolve_crate is a -1.28% improvement. If there's something going on here that needs fixing, I might need help figuring out what it is.

I did notice I was missing an important test though, so I've added that. (diff) @rustbot ready

oli-obk

sorry that took so long.

r=me with questions answered and nit applied

oli-obk · 2025-05-18T09:32:35Z

compiler/rustc_resolve/src/late.rs

        let mut bindings = smallvec![(PatBoundCtx::Product, Default::default())];
-        for (index, (pat, ty)) in inputs.enumerate() {
-            debug!(?pat, ?ty);
+        for (pat, _) in inputs.clone() {


Suggested change

for (pat, _) in inputs.clone() {

for pat in inputs.keys() {

this would require a bit more restructuring; inputs is an impl Iterator + Clone. it looks like in all instances it's from calling .iter().map(...) on a ThinVec. the closures passed to the map calls are cheap and pure, so it's probably not too bad perf-wise? but it looks like the pats are either always Some(_) or always None, so there is some room for optimization (assuming post-mono optimizations don't already do the loop simplification possible here. I haven't checked)

oli-obk · 2025-05-18T09:59:35Z

compiler/rustc_resolve/src/late.rs

-            this.fresh_binding(
-                Ident::new(kw::SelfLower, span),
-                delegation.id,
-                PatternSource::FnParam,
-                &mut bindings,
-            );
+            let ident = Ident::new(kw::SelfLower, span.normalize_to_macro_rules());
+            let res = Res::Local(delegation.id);
+            this.innermost_rib_bindings(ValueNS).insert(ident, res);


this is effectively just inlining the original fresh_binding impl and partially optimizing the inlined code for the case where bindings is Product with no recorded bindings, right? No actual changes

yeah, that's how I got that

oli-obk · 2025-05-18T10:01:22Z

compiler/rustc_resolve/src/late.rs

+                    self.resolve_pattern_inner(subpat, pat_src, bindings);
+                    // These bindings, but none from the surrounding pattern, are visible in the
+                    // guard; put them in scope and resolve `guard`.
+                    let subpat_bindings = bindings.pop().unwrap().1;


add an assert that the length of bindings is the same as the length of bindings before self.resolve_pattern_inner

done: (diff)

oli-obk · 2025-05-18T13:00:09Z

@bors r+ rollup

bors · 2025-05-18T13:00:11Z

📌 Commit f0b8ec1 has been approved by oli-obk

It is now in the queue for this repository.

name resolution for guard patterns This PR provides an initial implementation of name resolution for guard patterns [(RFC 3637)](https://github.com/rust-lang/rfcs/blob/master/text/3637-guard-patterns.md). This does not change the requirement that the bindings on either side of an or-pattern must be the same [(proposal here)](https://github.com/rust-lang/rfcs/blob/master/text/3637-guard-patterns.md#allowing-mismatching-bindings-when-possible); the code that handles that is separate from what this PR touches, so I'm saving it for a follow-up. On a technical level, this separates "collecting the bindings in a pattern" (which was already done for or-patterns) from "introducing those bindings into scope". I believe the approach used here can be extended straightforwardly in the future to work with `if let` guard patterns, but I haven't tried it myself since we don't allow those yet. Tracking issue for guard patterns: rust-lang#129967 cc `@Nadrieril`

Rollup of 10 pull requests Successful merges: - rust-lang#127013 (Add `f16` formatting and parsing) - rust-lang#138940 (Stabilize the avx512 target features) - rust-lang#140154 (Cygwin support in rustc) - rust-lang#140490 (split `asm!` parsing and validation) - rust-lang#140628 (std: stop using TLS in signal handler) - rust-lang#140746 (name resolution for guard patterns) - rust-lang#140926 (Return value of coroutine_layout fn changed to Result with LayoutError) - rust-lang#141127 (bump windows crate for compiler,bootstrap and tools) - rust-lang#141214 (Miri subtree update) - rust-lang#141218 (gvn: avoid creating overlapping assignments) r? `@ghost` `@rustbot` modify labels: rollup

Rollup of 8 pull requests Successful merges: - rust-lang#138940 (Stabilize the avx512 target features) - rust-lang#140490 (split `asm!` parsing and validation) - rust-lang#140628 (std: stop using TLS in signal handler) - rust-lang#140746 (name resolution for guard patterns) - rust-lang#140926 (Return value of coroutine_layout fn changed to Result with LayoutError) - rust-lang#141127 (bump windows crate for compiler,bootstrap and tools) - rust-lang#141214 (Miri subtree update) - rust-lang#141218 (gvn: avoid creating overlapping assignments) r? `@ghost` `@rustbot` modify labels: rollup

Rollup merge of rust-lang#140746 - dianne:guard-pat-res, r=oli-obk name resolution for guard patterns This PR provides an initial implementation of name resolution for guard patterns [(RFC 3637)](https://github.com/rust-lang/rfcs/blob/master/text/3637-guard-patterns.md). This does not change the requirement that the bindings on either side of an or-pattern must be the same [(proposal here)](https://github.com/rust-lang/rfcs/blob/master/text/3637-guard-patterns.md#allowing-mismatching-bindings-when-possible); the code that handles that is separate from what this PR touches, so I'm saving it for a follow-up. On a technical level, this separates "collecting the bindings in a pattern" (which was already done for or-patterns) from "introducing those bindings into scope". I believe the approach used here can be extended straightforwardly in the future to work with `if let` guard patterns, but I haven't tried it myself since we don't allow those yet. Tracking issue for guard patterns: rust-lang#129967 cc ``@Nadrieril``

dianne added 2 commits May 7, 2025 00:52

add a type alias for the pattern bindings stack

3007433

I'll be modifying it in future commits, so I think it's cleanest to abstract it out. Possibly a newtype would be ideal, but for now this is least disruptive.

rustbot assigned oli-obk May 7, 2025

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. labels May 7, 2025

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 May 8, 2025

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 May 8, 2025

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels May 8, 2025

dianne force-pushed the guard-pat-res branch from d7ac436 to 002e6e8 Compare May 9, 2025 06:21

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels May 9, 2025

oli-obk requested changes May 18, 2025

View reviewed changes

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels May 18, 2025

name resolution for guard patterns

f0b8ec1

dianne force-pushed the guard-pat-res branch from 002e6e8 to f0b8ec1 Compare May 18, 2025 11:22

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-author Status: This is awaiting some action (such as code changes or more information) from the author. labels May 18, 2025

fmease mentioned this pull request May 18, 2025

Rollup of 10 pull requests #141228

Closed

fmease mentioned this pull request May 18, 2025

Rollup of 8 pull requests #141232

Merged

bors merged commit 4e5b1aa into rust-lang:master May 18, 2025
6 checks passed

rustbot added this to the 1.89.0 milestone May 18, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

name resolution for guard patterns #140746

name resolution for guard patterns #140746

dianne commented May 7, 2025

rustbot commented May 7, 2025

Nadrieril commented May 8, 2025

This comment has been minimized.

bors commented May 8, 2025

bors commented May 8, 2025

This comment has been minimized.

rust-timer commented May 8, 2025

dianne commented May 8, 2025

rustbot commented May 8, 2025

dianne commented May 9, 2025

oli-obk left a comment

oli-obk May 18, 2025

dianne May 18, 2025

oli-obk May 18, 2025

dianne May 18, 2025

oli-obk May 18, 2025

dianne May 18, 2025

oli-obk commented May 18, 2025

bors commented May 18, 2025

name resolution for guard patterns #140746

name resolution for guard patterns #140746

Conversation

dianne commented May 7, 2025

rustbot commented May 7, 2025

Nadrieril commented May 8, 2025

This comment has been minimized.

bors commented May 8, 2025

bors commented May 8, 2025

This comment has been minimized.

rust-timer commented May 8, 2025

Overall result: ❌ regressions - please read the text below

dianne commented May 8, 2025

rustbot commented May 8, 2025

dianne commented May 9, 2025

oli-obk 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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

oli-obk commented May 18, 2025

bors commented May 18, 2025