Use `try_fold` in `Vec::extend_desugared` by el-ev · Pull Request #156440 · rust-lang/rust

el-ev · 2026-05-11T08:09:58Z

Try to revive #68046.
Resolves #63340.

Vec::extend_desugared is the fallback path for collect&extend from non-TrustedLen iterators. It previously used while let Some(element) = iterator.next(), which makes adapters like Chain to execute extra code on every call to next().

This PR switches to try_fold, which lets adapters use their specialized fold implementations, avoiding overhead.

rustbot · 2026-05-11T08:10:05Z

r? @Mark-Simulacrum

rustbot has assigned @Mark-Simulacrum.
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

Why was this reviewer chosen?

The reviewer was selected based on:

Owners of files modified in this PR: libs
libs expanded to 7 candidates
Random selection from Mark-Simulacrum, nia-e

el-ev · 2026-05-11T09:52:37Z

Tested locally:

bench_extend_chain_filter                                                                                                                                                    
                            main (ns)   HEAD (ns)   change                                                                                                                     
    0000_0010                  189.88      158.91     -16%                                                                                                                     
    0000_0100                  724.40      339.66     -53%                                                                                                                     
    0000_1000                 5106.20     1248.02     -76%                                                                                                                     
    0010_0010                  186.01      128.01     -31%                                                                                                                     
    0100_0100                  602.00      205.02     -66%                                                                                                                     
    1000_1000                 5251.13     1167.29     -78%
                                                                                                                                                                               
bench_collect_chain_filter                                      
                            main (ns)   HEAD (ns)   change                                                                                                                     
    0010                       142.93      137.68      -4%        
    0100                       512.12      305.18     -40%
    1000                      3004.60     1267.32     -58%                                                                                                                     
   
bench_extend_flatten                                                                                                                                                         
                            main (ns)   HEAD (ns)   change        
    0001_0010                  108.39      115.08      +6%
    0010_0100                  838.90      534.83     -36%                                                                                                                     
    0100_0100                 6745.96     1984.14     -71%
    1000_0010                 6196.60     1655.14     -73%

the8472 · 2026-05-11T18:13:36Z

Vec code is very hot and highly responsive to perturbations, let's see what perf says

@bors try @rust-timer queue

Use `try_fold` in `Vec::extend_desugared`

rust-bors · 2026-05-11T20:24:15Z

☀️ Try build successful (CI)
Build commit: dc931d4 (dc931d446f7059c1a67b165a4c202d6107484ba5, parent: 64a965e9013a9d14e83c4d370af26f6be6bf96fb)

rust-timer · 2026-05-11T21:05:08Z

Error occured while categorizing benchmark run:

could not find start commit for bound Commit("64a965e9013a9d14e83c4d370af26f6be6bf96fb")

mati865 · 2026-05-13T21:10:48Z

@rust-timer queue

mati865 · 2026-05-13T21:14:15Z

@rust-timer build dc931d4

rust-timer · 2026-05-13T21:14:17Z

Failed to enqueue some commit SHAs. Maybe they were already benchmarked?

mati865 · 2026-05-13T21:16:51Z

Scratch that. Let's start over.

@bors try @rust-timer queue

Use `try_fold` in `Vec::extend_desugared`

rust-bors · 2026-05-13T23:29:48Z

☀️ Try build successful (CI)
Build commit: 8902743 (8902743fec7a53b48dd90e2ff86517f69bd1fe0f, parent: 800892799d7666fe1dc17abd862100a6cf273718)

rust-timer · 2026-05-14T00:48:38Z

Finished benchmarking commit (8902743): comparison URL.

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

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

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=never
@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)	7.6%	[0.2%, 274.5%]	46
Regressions ❌ (secondary)	0.9%	[0.1%, 4.0%]	5
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.2%	[-0.3%, -0.0%]	16
All ❌✅ (primary)	7.6%	[0.2%, 274.5%]	46

Max RSS (memory usage)

Results (primary 2.8%, secondary -2.5%)

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

	mean	range	count
Regressions ❌ (primary)	3.6%	[0.8%, 14.2%]	7
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.7%	[-2.7%, -2.7%]	1
Improvements ✅ (secondary)	-2.5%	[-2.6%, -2.3%]	2
All ❌✅ (primary)	2.8%	[-2.7%, 14.2%]	8

Cycles

Results (primary 16.7%, secondary 3.7%)

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

	mean	range	count
Regressions ❌ (primary)	16.7%	[1.4%, 243.0%]	18
Regressions ❌ (secondary)	3.7%	[2.2%, 5.1%]	5
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	16.7%	[1.4%, 243.0%]	18

Binary size

Results (primary 1.2%, secondary 0.6%)

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

	mean	range	count
Regressions ❌ (primary)	1.2%	[0.0%, 4.0%]	60
Regressions ❌ (secondary)	0.6%	[0.0%, 1.0%]	9
Improvements ✅ (primary)	-0.2%	[-0.3%, -0.2%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.2%	[-0.3%, 4.0%]	62

Bootstrap: 507.838s -> 524.419s (3.27%)
Artifact size: 398.10 MiB -> 402.39 MiB (1.08%)

joboet

Also see #138752 for a prior attempt at doing this.

View changes since this review

joboet · 2026-05-21T15:28:20Z

-                self.set_len(len + 1);
+        let (lower, _) = iterator.size_hint();
+        self.reserve(lower);
+        while let Err(element) = iterator.try_fold((), |(), element| {


Vec::push_within_capacity does the same and avoids some unsafe.

hkBst · 2026-05-27T08:30:48Z

@joboet #138752 shows a similar +260% performance loss. Any idea what causes that?

Mark-Simulacrum · 2026-06-07T02:37:27Z

You can see in the diff of the codegen profiles (direct link) that something about this likely causes CGUs to shuffle such that the incr-println scenario now invalidates a bunch more LLVM CGUs, resulting in a ton more work.

I haven't yet had a chance to look at the implementation delta, but generally speaking I'm not particularly worried about the perf result there -- it's a fairly random effect most likely.

Mark-Simulacrum

Some more suggestions

View changes since this review

Mark-Simulacrum · 2026-06-21T16:07:42Z

+        let (lower, _) = iterator.size_hint();
+        self.reserve(lower);
+        while let Err(element) = iterator.try_fold((), |(), element| {
+            if self.len < self.capacity() {


Can we check the codegen for this? I'd expect it to help to make len and (probably) capacity local variables; I'd guess as-is LLVM can't tell the iterator and writing into the Vec won't mutate either.

This is probably a bit incompatible with the function suggestion above without a custom struct + fnmut impl (probably not worth it).

Mark-Simulacrum · 2026-06-21T16:09:13Z

+        let (lower, _) = iterator.size_hint();
+        self.reserve(lower);
+        while let Err(element) = iterator.try_fold((), |(), element| {
+            if self.len < self.capacity() {


Is there a reason we're changing the condition from equality to < here? It seems likely that equality gives more room for optimization.

Mark-Simulacrum · 2026-06-21T16:10:46Z

-                self.set_len(len + 1);
+        let (lower, _) = iterator.size_hint();
+        self.reserve(lower);
+        while let Err(element) = iterator.try_fold((), |(), element| {


I think we could have an inner function for this closure so that it is not instantiated for every passed iterator I, only every element type T. That should also help with the perf.rlo result I think?

The function may still want to be #[inline] (we should probably try both).

rustbot · 2026-06-21T16:14:30Z

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

Use try_fold in Vec::extend_desugared

278c76e

rustbot assigned Mark-Simulacrum May 11, 2026

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels May 11, 2026

el-ev marked this pull request as draft May 11, 2026 08:39

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 11, 2026

el-ev marked this pull request as ready for review May 11, 2026 09:52

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 11, 2026

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 11, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request May 11, 2026

Auto merge of #156440 - el-ev:perf_collect, r=<try>

dc931d4

Use `try_fold` in `Vec::extend_desugared`

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request May 13, 2026

Auto merge of #156440 - el-ev:perf_collect, r=<try>

8902743

Use `try_fold` in `Vec::extend_desugared`

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 14, 2026

joboet reviewed May 21, 2026

View reviewed changes

Mark-Simulacrum requested changes Jun 21, 2026

View reviewed changes

rustbot removed the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jun 21, 2026

rustbot added the S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. label Jun 21, 2026

Uh oh!

Uh oh!

Conversation

el-ev commented May 11, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented May 11, 2026

Uh oh!

el-ev commented May 11, 2026

Uh oh!

the8472 commented May 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment has been minimized.

This comment has been minimized.

rust-bors Bot commented May 11, 2026

Uh oh!

This comment has been minimized.

rust-timer commented May 11, 2026

Uh oh!

mati865 commented May 13, 2026

Uh oh!

This comment has been minimized.

mati865 commented May 13, 2026

Uh oh!

rust-timer commented May 13, 2026

Uh oh!

mati865 commented May 13, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

rust-bors Bot commented May 13, 2026

Uh oh!

This comment has been minimized.

rust-timer commented May 14, 2026

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

Uh oh!

joboet 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!

hkBst commented May 27, 2026

Uh oh!

Mark-Simulacrum commented Jun 7, 2026

Uh oh!

Mark-Simulacrum 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!

rustbot commented Jun 21, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

el-ev commented May 11, 2026 •

edited by rustbot

Loading

the8472 commented May 11, 2026 •

edited

Loading

joboet left a comment •

edited by rustbot

Loading

Mark-Simulacrum left a comment •

edited by rustbot

Loading