perf: reduce module concatenation hot-path overhead

[LingyuCoder]

Summary

• Share unsupported-syntax bailout payloads instead of cloning their module lists.
• Use cached-key sorting for repeated readable identifiers.
• Clone root chunk state once per root search instead of once per candidate.

Related links

• Split from perf: optimize module concatenation root search #14815.
• Stack 4/4; based on perf: reuse module concatenation search state #14833.

Checklist

• Tests updated in the preceding stack layer.
• Documentation not required.

[github-actions]

📝 Benchmark detail: Open

Name	Base (25274d8)	Current	Change
10000_big_production-mode_disable-minimize + exec	12.1 s ± 334 ms	12.1 s ± 352 ms	-0.14 %
10000_development-mode + exec	895 ms ± 19 ms	905 ms ± 93 ms	+1.12 %
10000_development-mode_hmr + stats	162 ms ± 18 ms	152 ms ± 3.4 ms	-5.92 %
10000_development-mode_noop-loader + exec	1.78 s ± 135 ms	1.75 s ± 157 ms	-1.80 %
10000_production-mode + exec	1.02 s ± 32 ms	1.02 s ± 26 ms	-0.43 %
10000_production-mode_persistent-cold + exec	1.21 s ± 38 ms	1.17 s ± 42 ms	-2.77 %
10000_production-mode_persistent-hot + exec	634 ms ± 12 ms	628 ms ± 8.2 ms	-1.07 %
10000_production-mode_source-map + exec	1.21 s ± 97 ms	1.17 s ± 58 ms	-3.54 %
arco-pro_development-mode + exec	1.25 s ± 40 ms	1.3 s ± 94 ms	+4.10 %
arco-pro_development-mode_hmr + stats	33 ms ± 1.3 ms	33 ms ± 0.63 ms	-1.86 %
arco-pro_production-mode + exec	2.34 s ± 154 ms	2.32 s ± 116 ms	-0.77 %
arco-pro_production-mode_generate-package-json-webpack-plugin + exec	2.42 s ± 84 ms	2.36 s ± 83 ms	-2.50 %
arco-pro_production-mode_persistent-cold + exec	2.4 s ± 129 ms	2.39 s ± 65 ms	-0.67 %
arco-pro_production-mode_persistent-hot + exec	344 ms ± 9 ms	339 ms ± 8.8 ms	-1.28 %
arco-pro_production-mode_source-map + exec	2.84 s ± 68 ms	2.86 s ± 194 ms	+0.56 %
arco-pro_production-mode_source-map_persistent-cold + exec	2.95 s ± 129 ms	2.87 s ± 104 ms	-2.50 %
arco-pro_production-mode_source-map_persistent-hot + exec	477 ms ± 7.1 ms	483 ms ± 9.9 ms	+1.17 %
arco-pro_production-mode_traverse-chunk-modules + exec	2.37 s ± 143 ms	2.36 s ± 71 ms	-0.50 %
bundled-threejs_development-mode + exec	181 ms ± 2.9 ms	181 ms ± 1.2 ms	+0.11 %
bundled-threejs_production-mode + exec	208 ms ± 5.7 ms	208 ms ± 6.5 ms	-0.10 %
large-dyn-imports_development-mode + exec	1.17 s ± 31 ms	1.17 s ± 27 ms	+0.41 %
large-dyn-imports_production-mode + exec	1.27 s ± 30 ms	1.26 s ± 32 ms	-0.77 %
threejs_development-mode_10x + exec	775 ms ± 13 ms	772 ms ± 31 ms	-0.39 %
threejs_development-mode_10x_hmr + stats	112 ms ± 2.7 ms	110 ms ± 20 ms	-1.81 %
threejs_production-mode_10x + exec	2.86 s ± 136 ms	2.81 s ± 42 ms	-1.67 %
threejs_production-mode_10x_persistent-cold + exec	2.95 s ± 26 ms	2.91 s ± 29 ms	-1.29 %
threejs_production-mode_10x_persistent-hot + exec	380 ms ± 7.9 ms	377 ms ± 11 ms	-0.68 %
threejs_production-mode_10x_source-map + exec	3.66 s ± 179 ms	3.62 s ± 183 ms	-1.24 %
10000_big_production-mode_disable-minimize + rss memory	2179 MiB ± 83.1 MiB	2217 MiB ± 23.4 MiB	+1.79 %
10000_development-mode + rss memory	538 MiB ± 14 MiB	543 MiB ± 11.3 MiB	+1.05 %
10000_development-mode_hmr + rss memory	788 MiB ± 40.7 MiB	794 MiB ± 30.1 MiB	+0.88 %
10000_development-mode_noop-loader + rss memory	805 MiB ± 11.9 MiB	812 MiB ± 7.42 MiB	+0.90 %
10000_production-mode + rss memory	466 MiB ± 9.03 MiB	471 MiB ± 10.9 MiB	+1.05 %
10000_production-mode_persistent-cold + rss memory	662 MiB ± 11.1 MiB	665 MiB ± 11.5 MiB	+0.37 %
10000_production-mode_persistent-hot + rss memory	649 MiB ± 33.8 MiB	666 MiB ± 24.7 MiB	+2.68 %
10000_production-mode_source-map + rss memory	483 MiB ± 32.9 MiB	471 MiB ± 59 MiB	-2.53 %
arco-pro_development-mode + rss memory	399 MiB ± 2.04 MiB	406 MiB ± 6.39 MiB	+1.66 %
arco-pro_development-mode_hmr + rss memory	422 MiB ± 8.17 MiB	432 MiB ± 15 MiB	+2.41 %
arco-pro_production-mode + rss memory	574 MiB ± 12.9 MiB	581 MiB ± 8.31 MiB	+1.18 %
arco-pro_production-mode_generate-package-json-webpack-plugin + rss memory	588 MiB ± 8.89 MiB	593 MiB ± 14.2 MiB	+0.89 %
arco-pro_production-mode_persistent-cold + rss memory	660 MiB ± 26.3 MiB	661 MiB ± 11.9 MiB	+0.16 %
arco-pro_production-mode_persistent-hot + rss memory	293 MiB ± 9.15 MiB	293 MiB ± 3.64 MiB	-0.21 %
arco-pro_production-mode_source-map + rss memory	682 MiB ± 11.2 MiB	682 MiB ± 6.28 MiB	-0.05 %
arco-pro_production-mode_source-map_persistent-cold + rss memory	812 MiB ± 37.6 MiB	811 MiB ± 24 MiB	-0.08 %
arco-pro_production-mode_source-map_persistent-hot + rss memory	413 MiB ± 12.3 MiB	426 MiB ± 12 MiB	+3.19 %
arco-pro_production-mode_traverse-chunk-modules + rss memory	579 MiB ± 11.1 MiB	581 MiB ± 10.3 MiB	+0.28 %
bundled-threejs_development-mode + rss memory	126 MiB ± 3.29 MiB	129 MiB ± 7.42 MiB	+2.18 %
bundled-threejs_production-mode + rss memory	137 MiB ± 2.18 MiB	139 MiB ± 4.97 MiB	+1.37 %
large-dyn-imports_development-mode + rss memory	553 MiB ± 20.4 MiB	553 MiB ± 17.5 MiB	+0.02 %
large-dyn-imports_production-mode + rss memory	406 MiB ± 1.63 MiB	406 MiB ± 2.91 MiB	+0.05 %
threejs_development-mode_10x + rss memory	432 MiB ± 9.74 MiB	435 MiB ± 4.68 MiB	+0.83 %
threejs_development-mode_10x_hmr + rss memory	592 MiB ± 10.9 MiB	608 MiB ± 18.5 MiB	+2.76 %
threejs_production-mode_10x + rss memory	615 MiB ± 22.8 MiB	616 MiB ± 23.4 MiB	+0.11 %
threejs_production-mode_10x_persistent-cold + rss memory	576 MiB ± 12 MiB	571 MiB ± 8.94 MiB	-1.01 %
threejs_production-mode_10x_persistent-hot + rss memory	378 MiB ± 7.36 MiB	375 MiB ± 10.3 MiB	-0.69 %
threejs_production-mode_10x_source-map + rss memory	738 MiB ± 26.3 MiB	736 MiB ± 28.2 MiB	-0.24 %

Base persistent cache hit rate: 👍

Current persistent cache hit rate: 👍

[github-actions]

📦 Binary Size-limit

Comparing 7cc987b to Merge 5bd606d24afa23162dac7ae8397a2295ba1d39b9 into 7d705ba6b17e919e1ea703aaa8278e1b2efa59ae by harpsealjs

⏳ The base commit triggered a linux binding build, but its binary size data has not been generated yet, so the size comparison is skipped.

Please re-run this workflow once the ecosystem-benchmark data for that commit is published.

Warning

Reference only — not the real baseline. The base commit's data isn't ready yet, so this compares against the nearest earlier commit that has data (5cfe1be) for a rough estimate:

❌ Size increased by 8.00KB from 66.66MB to 66.67MB (⬆️0.01%)

[github-actions]

Rsdoctor Bundle Diff Analysis

Found 5 projects in monorepo, 0 projects with changes.

📊 Quick Summary

Project	Total Size	Gzip Size	Change	Gzip Change
react-10k	5.6 MB	1.3 MB	0	0
popular-libs	1.7 MB	552.1 KB	0	0
react-1k	823.1 KB	217.0 KB	0	0
react-5k	2.7 MB	663.8 KB	0	0
ui-components	4.9 MB	1.4 MB	0	0

Generated by Rsdoctor GitHub Action

[codspeed-hq]

Merging this PR will degrade performance by 0.41%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 1 improved benchmark
❌ 1 regressed benchmark
✅ 44 untouched benchmarks
⏩ 47 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	rust@create_chunk_assets	1.4 ms	1.5 ms	-3.02%
⚡	Simulation	rust@create_concatenate_module_unsupported_syntax	18.1 ms	17.7 ms	+2.27%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing seal/module-concat-hot-paths (7cc987b) with seal/module-concat-search-state (5bd606d)}

Footnotes

1. 47 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[LingyuCoder]

@codspeedbot explain the regression

[codspeed-hq]

@codspeedbot explain the regression

rust@create_chunk_assets regression: environment artifact, not a real regression

Short version: the flagged -3.02% (1.4 ms → 1.5 ms) on rust@create_chunk_assets is a measurement artifact from base and head running in different runtime environments, not an effect of this PR's code.

Evidence

1. Base and head ran on different runners. The comparison detected these environment differences, and create_chunk_assets is explicitly among the benchmarks whose base and head crossed them:

• CPU: AMD EPYC 9V74 80-Core (base) → AMD EPYC 7763 64-Core (head)
• linked library libm.so.6 differs between the two runs

In Simulation mode the CPU cycle estimate is derived from a cache model tied to the physical CPU, so a different CPU/toolchain shifts the estimate on byte-identical code. That's exactly what CodSpeed's "Different runtime environments detected" warning on the report is flagging.

2. This PR does not touch the create_chunk_assets code path. The changes are confined to module concatenation (crates/rspack_plugin_javascript/src/plugin/module_concatenation_plugin.rs — the shared unsupported-syntax payloads, cached-key sorting, and per-root chunk-state clone). The create_chunk_assets hot path is output-filename template rendering in rspack_core (render_manifest → render_template → placeholder replacement), which none of these changes reach.

3. The benchmark is short and codegen-sensitive. create_chunk_assets is a ~1.4 ms Simulation stage bench; sub-~4% swings on it are routinely codegen/inlining/layout artifacts of the shared bench binary rather than real work changes. A 1.4 ms → 1.5 ms swing that only appears across a CPU + libm change is consistent with that.

On the flip side

The paired ⚡ +32.87% on sources@parse_source_map_from_json (9.9 ms → 7.5 ms) crossed the same environment difference, so it is not a real improvement from this PR either — it's the mirror image of the same cross-runner effect.

Recommendation

No source-level fix targets this — there's nothing in the module-concatenation change to correct. It's safe to acknowledge this regression on CodSpeed. If you'd like a clean signal, re-run once so base and head land on the same runner CPU; the real signal for this PR is the concatenation benchmarks (rust@create_concatenate_module_*), which are the ones this change actually affects.

Summary

• Warming up the performance wizard...
• Handling comment — Comment handling finished successfully

_{Was this response helpful? React with 👍 or 👎 to help us improve CodSpeedBot.}