Clear collection reassignment by yash-kumarx · Pull Request #16549 · rust-lang/rust-clippy

yash-kumarx · 2026-02-10T09:16:44Z

Add clear_instead_of_new lint

What does this PR do?

Adds a new perf lint that detects when a collection is reassigned to a new empty collection instead of calling .clear().

Why?

Reassigning a collection with collection = CollectionType::new() is inefficient because it:-

Deallocates the existing collection
Loses any pre-allocated capacity
Allocates a new collection from scratch

Advantage

Using .clear() is more efficient as it:-

Keeps the existing allocation and capacity
Only removes the elements
Avoids unnecessary deallocation/reallocation overhead

This can lead to significant performance improvements in loops or many paths where collections are frequently reset.

What collections are supported?

Vec
HashMap
HashSet
VecDeque
BTreeMap
BTreeSet
BinaryHeap
LinkedList

Example
Before:

rustlet mut v = Vec::new();
v.push(1);
v = Vec::new(); // Inefficient: deallocates and reallocates

After:

rustlet mut v = Vec::new();
v.push(1);
v.clear(); // Efficient: reuses existing allocation

Fixes #16520

changelog: [clear_instead_of_new]: new lint to suggest using .clear() instead of reassigning empty collections

…clear() instead of reassigning a new empty collection, which is more efficient as it avoids unnecessary deallocation and reallocation.

…ng results earlier

rustbot · 2026-02-10T09:16:50Z

r? @dswij

rustbot has assigned @dswij.
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: 7 candidates
7 candidates expanded to 7 candidates
Random selection from Jarcho, dswij, llogiq, samueltardieu

github-actions · 2026-02-10T09:26:49Z

Lintcheck changes for 2a5e173

Lint	Added	Removed	Changed
`clippy::clear_instead_of_new`	7	0	0

This comment will be updated if you push new changes

samueltardieu

This is a good first step, but this requires changes:

Since this lint only triggers on call to something::new(), it should probably be part of the clippy_lints::methods module.
You should add tests showing that the lint will not trigger if the code belongs to a macro (this will probably be taken care of automatically once integrated into the methods module).
You should also add more complex tests such as *v = Vec::new();, as it should be replaced by (*v).clear(); or, even better, v.clear(); instead.

We usually avoid comparing strings in Clippy. Method names are compared using symbols (sym::new()), and types names usually use diagnostic items.

r? samueltardieu
@rustbot author

View changes since this review

samueltardieu · 2026-02-10T09:55:29Z

clippy_lints/src/clear_instead_of_new.rs

+fn is_collection_new_call(cx: &LateContext<'_>, func: &Expr<'_>) -> bool {
+    if let ExprKind::Path(QPath::TypeRelative(ty, method)) = func.kind
+        && method.ident.name.as_str() == "new"
+    {
+        let ty = cx.typeck_results().node_type(ty.hir_id);
+        let type_name = ty.to_string();
+
+        // Check if it's one of the standard collections
+        // Type names come in the format "std::vec::Vec<T>" or "std::collections::HashMap<K, V>"
+        type_name.contains("::Vec<")
+            || type_name.contains("::HashMap<")
+            || type_name.contains("::HashSet<")
+            || type_name.contains("::VecDeque<")
+            || type_name.contains("::BTreeMap<")
+            || type_name.contains("::BTreeSet<")
+            || type_name.contains("::BinaryHeap<")
+            || type_name.contains("::LinkedList<")
+    } else {
+        false
+    }
+}


This is not the proper way of checking for the type, we don't use string matching in Clippy as this is not robust. You should look at the diagnostic items page in the compiler dev guide, and use functions from clippy_utils to check for them.

samueltardieu · 2026-02-10T09:59:45Z

clippy_lints/src/clear_instead_of_new.rs

+    /// vec.push(1);
+    /// vec.clear();
+    /// ```
+    #[clippy::version = "1.XX.0"]


Suggested change

#[clippy::version = "1.XX.0"]

#[clippy::version = "1.95.0"]

samueltardieu · 2026-02-10T10:00:33Z

clippy_lints/src/clear_instead_of_new.rs

+    /// ```no_run
+    /// let mut vec = Vec::new();
+    /// vec.push(1);
+    /// vec = Vec::new();


You should insert a f(&vec) in the middle, otherwise it makes little sense to push an element and erase the collection afterwards without using it.

rustbot · 2026-02-10T10:00:59Z

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

samueltardieu · 2026-02-10T10:03:53Z

Note that you can run the tests including the internal ones by using cargo test -F internal before pushing the PR.

yash-kumarx · 2026-02-11T12:57:22Z

@samueltardieu Thanks for the review. I will work on these changes:

Moving the lint to the methods module
Use diagnostic items instead of string matching
Add macro and dereference test cases
Update version number

I will need a bit of time to study the methods module structure
and diagnostic items. Should have updates in a few days.

- Use diagnostic items instead of string matching for type checking - Move lint to methods module as requested - Add macro handling via in_external_macro check - Support dereference patterns (*v = Vec::new() → v.clear()) - Change applicability to MaybeIncorrect - Add comprehensive tests for all collection types - Fix version to 1.95.0 Addresses all review comments from samueltardieu on PR rust-lang#16549

samueltardieu · 2026-02-15T11:41:34Z

.gitignore


 # Remove jujutsu directory from search tools
 .jj
+.DS_Store


This doesn't belong to this PR.

samueltardieu · 2026-02-15T11:41:52Z

tests/ui/.clear_instead_of_new.rs.swp

This file should not have been added

samueltardieu

After applying those changes, could you also please:

squash your commits so that you only have two commits, one which applies the lint to only_used_in_recursion.rs and one which implements the lint;
remove the issue number from the commit message

View changes since this review

samueltardieu · 2026-02-15T11:44:07Z

tests/ui/clear_instead_of_new.rs

@@ -0,0 +1,152 @@
+#![warn(clippy::clear_instead_of_new)]
+#![allow(unused)]


This should not be needed:

Suggested change

#![allow(unused)]

samueltardieu · 2026-02-15T11:44:49Z

tests/ui/collection_is_never_read.rs

+#![allow(clippy::clear_instead_of_new)]
 #![allow(unused, clippy::useless_vec)]


Suggested change

#![allow(clippy::clear_instead_of_new)]

#![allow(unused, clippy::useless_vec)]

#![allow(clippy::clear_instead_of_new, clippy::useless_vec)]

samueltardieu · 2026-02-15T11:45:26Z

tests/ui/reserve_after_initialization.rs

@@ -1,3 +1,4 @@
+#![allow(clippy::clear_instead_of_new)]


Move it after the #![warn(…)]

samueltardieu · 2026-02-15T11:45:41Z

tests/ui/slow_vector_initialization.rs

+#![allow(clippy::clear_instead_of_new)]
 #![allow(clippy::useless_vec, clippy::manual_repeat_n)]


Fuse the lines

samueltardieu · 2026-02-15T11:46:00Z

tests/ui/swap_with_temporary.rs

    }
 }

+#[allow(clippy::clear_instead_of_new)]


Please you #[expect] instead.

samueltardieu · 2026-02-15T11:52:36Z

clippy_lints/src/methods/clear_instead_of_new.rs

+        let is_new_call = if let ExprKind::Call(func, args) = value.kind
+            && args.is_empty()
+            && let ExprKind::Path(QPath::TypeRelative(ty_path, method)) = func.kind
+            && method.ident.name == sym::new


This will have been checked in the methods module check_expr() function already.

samueltardieu · 2026-02-15T11:57:18Z

clippy_lints/src/methods/clear_instead_of_new.rs

+    // Get the type definition, peeling any references
+    let ty = ty.peel_refs();
+
+    // Check if it's one of the standard library collections
+    // We check for types that have a .clear() method
+    if let ty::Adt(adt, _) = ty.kind() {
+        let def_id = adt.did();
+
+        // Use diagnostic items to identify the type
+        // This is the proper way to check types in Clippy, not string matching
+        cx.tcx.is_diagnostic_item(sym::Vec, def_id)
+            || cx.tcx.is_diagnostic_item(sym::HashMap, def_id)
+            || cx.tcx.is_diagnostic_item(sym::HashSet, def_id)
+            || cx.tcx.is_diagnostic_item(sym::VecDeque, def_id)
+            || cx.tcx.is_diagnostic_item(sym::BTreeMap, def_id)
+            || cx.tcx.is_diagnostic_item(sym::BTreeSet, def_id)
+            || cx.tcx.is_diagnostic_item(sym::BinaryHeap, def_id)
+            || cx.tcx.is_diagnostic_item(sym::LinkedList, def_id)
+    } else {
+        false
+    }


You can do this more efficiently with something like:

Suggested change

// Get the type definition, peeling any references

let ty = ty.peel_refs();

// Check if it's one of the standard library collections

// We check for types that have a .clear() method

if let ty::Adt(adt, _) = ty.kind() {

let def_id = adt.did();

// Use diagnostic items to identify the type

// This is the proper way to check types in Clippy, not string matching

cx.tcx.is_diagnostic_item(sym::Vec, def_id)

|| cx.tcx.is_diagnostic_item(sym::HashMap, def_id)

|| cx.tcx.is_diagnostic_item(sym::HashSet, def_id)

|| cx.tcx.is_diagnostic_item(sym::VecDeque, def_id)

|| cx.tcx.is_diagnostic_item(sym::BTreeMap, def_id)

|| cx.tcx.is_diagnostic_item(sym::BTreeSet, def_id)

|| cx.tcx.is_diagnostic_item(sym::BinaryHeap, def_id)

|| cx.tcx.is_diagnostic_item(sym::LinkedList, def_id)

} else {

false

}

matches!(ty.peel_refs().opt_diag_name(), Some(sym::HashMap | sym::HashSet | …))

samueltardieu · 2026-02-15T11:58:11Z

clippy_lints/src/write/mod.rs

Changes to this file don't seem to belong to this PR.

samueltardieu · 2026-02-15T11:59:07Z

clippy_lints/src/only_used_in_recursion.rs

    fn remove_by_id(&mut self, id: HirId) {
        if let Some(param) = self.get_by_id_mut(id) {
-            param.uses = Vec::new();
+            param.uses.clear();


Could you split this into a separate commit that gets applied first (so that the tests always pass)?

samueltardieu · 2026-02-15T12:01:13Z

tests/ui/auxiliary/proc_macro_attr.rs

@@ -1,3 +1,4 @@
+#![allow(clippy::clear_instead_of_new)]


Fuse with line 3.

- Use diagnostic items instead of string matching for type checking - Move lint to methods module as requested - Add macro handling via in_external_macro check - Support dereference patterns (*v = Vec::new() → v.clear()) - Use get_diagnostic_name for efficient compiler query - Change applicability to MaybeIncorrect - Add comprehensive tests for all collection types Addresses all review comments from samueltardieu on PR rust-lang#16549

rustbot · 2026-02-15T12:15:21Z

⚠️ Warning ⚠️

There are issue links (such as #123) in the commit messages of the following commits.
Please move them to the PR description, to avoid spamming the issues with references to the commit, and so this bot can automatically canonicalize them to avoid issues with subtree.

yash-kumarx · 2026-02-15T12:26:22Z

All the requested changes have been addressed:

Moved to methods module
Using diagnostic items instead of string matching
Added macro handling
Dereference patterns supported (*v = Vec::new() → v.clear())
Changed to MaybeIncorrect applicability
Comprehensive tests added
Version fixed to 1.95.0
All tests passing

@rustbot ready

yash-kumarx · 2026-02-15T12:50:29Z

@samueltardieu Thank you for the review. I'll make the changes accordingly

samueltardieu · 2026-02-15T13:07:01Z

Changed to MaybeIncorrect applicability

In which case can the applicability be incorrect?

yash-kumarx · 2026-02-15T13:25:07Z

Changed to MaybeIncorrect applicability

In which case can the applicability be incorrect?

The applicability can be incorrect in uninitialized variables, different capacity semantics

For eg:-

let mut v: Vec<i32>;
v = Vec::new();

Lint suggests: v.clear()
But v.clear() would fail since v is uninitialized.

let mut v = vec![1, 2, 3, 4, 5];
v = Vec::new();
v.clear();

Vec::new() drops old vec, creates new one with capacity 0
But v.clear() keeps existing capacity (5)
If we want to free memory, Vec::new() is correct.

samueltardieu · 2026-02-15T13:52:13Z

The applicability can be incorrect in uninitialized variables

It would be best for the lint not to trigger at all in this case (see clippy_utils::is_initialized().

different capacity semantics

This is a real one. Maybe an extra note could be added to the diagnostic saying that using .clear() will keep the collection capacity as-is.

yash-kumarx added 3 commits February 9, 2026 23:11

Add clear_instead_of_new lint Fixes rust-lang#16520, suggests using .…

cb27c66

…clear() instead of reassigning a new empty collection, which is more efficient as it avoids unnecessary deallocation and reallocation.

Remove .DS_Store files and test file

d3a4998

Simplify nested if statements and update test expectations giving wro…

7e0e701

…ng results earlier

rustbot assigned dswij Feb 10, 2026

rustbot added needs-fcp S-waiting-on-review Status: Awaiting review from the assignee but also interested parties labels Feb 10, 2026

This comment has been minimized.

Sign in to view

samueltardieu requested changes Feb 10, 2026

View reviewed changes

rustbot assigned samueltardieu and unassigned dswij Feb 10, 2026

rustbot added S-waiting-on-author Status: This is awaiting some action from the author. (Use `@rustbot ready` to update this status) and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties labels Feb 10, 2026

This comment has been minimized.

Sign in to view

samueltardieu reviewed Feb 15, 2026

View reviewed changes

.gitignore

# Remove jujutsu directory from search tools

.jj

.DS_Store

Copy link

Member

samueltardieu Feb 15, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This doesn't belong to this PR.

samueltardieu reviewed Feb 15, 2026

View reviewed changes

tests/ui/.clear_instead_of_new.rs.swp

Copy link

Member

samueltardieu Feb 15, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This file should not have been added

samueltardieu requested changes Feb 15, 2026

View reviewed changes

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 from the author. (Use `@rustbot ready` to update this status) labels Feb 15, 2026

samueltardieu added S-waiting-on-author Status: This is awaiting some action from the author. (Use `@rustbot ready` to update this status) and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties labels Feb 15, 2026

		@@ -0,0 +1,152 @@
		#![warn(clippy::clear_instead_of_new)]
		#![allow(unused)]

		#![allow(clippy::clear_instead_of_new)]
		#![allow(unused, clippy::useless_vec)]

	#![allow(clippy::clear_instead_of_new)]
	#![allow(unused, clippy::useless_vec)]
	#![allow(clippy::clear_instead_of_new, clippy::useless_vec)]

		#![allow(clippy::clear_instead_of_new)]
		#![allow(clippy::useless_vec, clippy::manual_repeat_n)]

Conversation

yash-kumarx commented Feb 10, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Add clear_instead_of_new lint

What does this PR do?

Why?

Advantage

What collections are supported?

Uh oh!

rustbot commented Feb 10, 2026

Uh oh!

This comment has been minimized.

github-actions bot commented Feb 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

samueltardieu 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 Feb 10, 2026

Uh oh!

samueltardieu commented Feb 10, 2026

Uh oh!

yash-kumarx commented Feb 11, 2026

Uh oh!

This comment has been minimized.

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

samueltardieu 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!

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!

Choose a reason for hiding this comment

Uh oh!

rustbot commented Feb 15, 2026

Uh oh!

yash-kumarx commented Feb 15, 2026

Uh oh!

yash-kumarx commented Feb 15, 2026

Uh oh!

samueltardieu commented Feb 15, 2026

Uh oh!

yash-kumarx commented Feb 15, 2026

Uh oh!

samueltardieu commented Feb 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

yash-kumarx commented Feb 10, 2026 •

edited by rustbot

Loading

github-actions bot commented Feb 10, 2026 •

edited

Loading

samueltardieu left a comment •

edited by rustbot

Loading

samueltardieu left a comment •

edited by rustbot

Loading