Optimization in use_reducer

[Pascal-So]

Description

Yew currently provides two versions of the reducer hook:

• use_reducer: This hook always causes a re-render when an action is dispatched.
• use_reducer_eq: This hook avoids the re-render if the new state is the same as the old state. However, the state now has to implement PartialEq.

I feel like there is room for a middle ground: The reducer takes an Rc and returns an Rc. Even if we don't have a PartialEq implementation on the state, we can still know that the state must be unchanged if the reducer returns the exact same Rc back to us. This check is almost free (just a pointer comparison).

This PR adds a simple check if the two Rcs are the same, and in that case skip the re-render.

Note that even for use_reducer_eq this check is beneficial, since the PartialEq implementation might be more expensive than just a pointer comparison.

This change might be a breaking change if someone was relying on the exact number of re-renders, but I'd assume that this is not covered in Yew's semver guarantees, right?

Checklist

• [x] I have reviewed my own code
• [x] I have added tests

[github-actions[bot]]

Visit the preview URL for this PR (updated for commit 5e3dadf):

https://yew-rs-api--pr3945-use-reducer-optimiza-u4xs77n3.web.app

_{(expires Tue, 25 Nov 2025 22:37:31 GMT)}

_{🔥 via Firebase Hosting GitHub Action 🌎}

[github-actions[bot]]

Benchmark - core

Yew Master

vnode           fastest       │ slowest       │ median        │ mean          │ samples │ iters
╰─ vnode_clone  2.67 ns       │ 3.139 ns      │ 2.675 ns      │ 2.684 ns      │ 100     │ 1000000000

Pull Request

vnode           fastest       │ slowest       │ median        │ mean          │ samples │ iters
╰─ vnode_clone  2.712 ns      │ 3.025 ns      │ 2.717 ns      │ 2.721 ns      │ 100     │ 1000000000

[github-actions[bot]]

Benchmark - SSR

Yew Master

Benchmark	Round	Min (ms)	Max (ms)	Mean (ms)	Standard Deviation
Baseline	10	270.214	272.125	270.561	0.565
Hello World	10	470.772	482.678	478.514	3.807
Function Router	10	1550.812	1586.253	1568.509	10.931
Concurrent Task	10	1005.011	1006.697	1005.780	0.588
Many Providers	10	1031.203	1065.500	1049.301	11.537

Pull Request

Benchmark	Round	Min (ms)	Max (ms)	Mean (ms)	Standard Deviation
Baseline	10	281.405	281.765	281.542	0.116
Hello World	10	477.791	496.804	481.093	5.609
Function Router	10	1602.130	1617.413	1607.869	3.923
Concurrent Task	10	1004.972	1007.096	1006.098	0.810
Many Providers	10	1057.383	1115.522	1079.133	15.417

[github-actions[bot]]

Size Comparison

examples	master (KB)	pull request (KB)	diff (KB)	diff (%)
async_clock	98.312	98.312	0	0.000%
boids	166.434	166.434	0	0.000%
communication_child_to_parent	91.440	91.440	0	0.000%
communication_grandchild_with_grandparent	102.572	102.572	0	0.000%
communication_grandparent_to_grandchild	98.954	98.954	0	0.000%
communication_parent_to_child	88.811	88.811	0	0.000%
contexts	103.940	103.943	+0.003	+0.003%
counter	85.489	85.489	0	0.000%
counter_functional	85.756	85.759	+0.003	+0.003%
dyn_create_destroy_apps	88.647	88.647	0	0.000%
file_upload	97.972	97.972	0	0.000%
function_delayed_input	91.063	91.066	+0.003	+0.003%
function_memory_game	169.376	169.381	+0.005	+0.003%
function_router	328.280	328.325	+0.045	+0.014%
function_todomvc	161.585	161.599	+0.014	+0.008%
futures	236.928	236.928	0	0.000%
game_of_life	103.849	103.849	0	0.000%
immutable	192.977	192.979	+0.003	+0.002%
inner_html	79.966	79.966	0	0.000%
js_callback	107.514	107.517	+0.003	+0.003%
keyed_list	179.134	179.134	0	0.000%
mount_point	83.186	83.186	0	0.000%
nested_list	112.699	112.699	0	0.000%
node_refs	90.833	90.833	0	0.000%
password_strength	1743.419	1743.419	0	0.000%
portals	92.307	92.307	0	0.000%
router	301.961	301.967	+0.006	+0.002%
suspense	111.686	111.688	+0.003	+0.003%
timer	88.241	88.241	0	0.000%
timer_functional	95.934	95.938	+0.005	+0.005%
todomvc	141.684	141.684	0	0.000%
two_apps	85.341	85.341	0	0.000%
web_worker_fib	132.882	132.896	+0.014	+0.010%
web_worker_prime	184.158	184.185	+0.026	+0.014%
webgl	82.524	82.524	0	0.000%

✅ None of the examples has changed their size significantly.