perf: memoize plugin loader alias maps by amknight · Pull Request #69316 · openclaw/openclaw

amknight · 2026-04-20T11:46:16Z

Summary

memoize plugin loader alias maps by effective resolution context
reuse the resolved Jiti loader config/cache key on hot loader-cache hits
bound the loader alias/config caches to avoid unbounded growth on unique module paths
keep cache keys sensitive to cwd, production mode, and private QA alias enablement

Why

OpenClaw hits this in real plugin/channel loader paths. A cached Jiti loader hit still had to rebuild the plugin SDK alias map and stringify the alias map into a Jiti cache key before it could discover that the loader was already cached.

Reading local jiti@2.6.1 source showed that this is not about reusing Jiti's internal alias sentinel across fresh createJiti() calls. Jiti normalizes aliases into its own internal object. The useful optimization is avoiding repeated OpenClaw alias/config resolution before returning an already-cached loader.

Measurements

Local synthetic measurements on this machine; medians across repeated samples.

Benchmark	main	first PR patch	refined PR
`buildPluginLoaderAliasMap()` same module, 5,000 calls	3676.79ms	4.74ms	4.74ms
`getCachedPluginJitiLoader()` same module, 5,000 calls	13770.14ms	1392.08ms	41.71ms
`loadChannelPluginModule()` same module, 5,000 hot loads	17174.67ms	3791.66ms	583.18ms

Tradeoff

This helps repeated same-context loader resolution: long-running processes, channel/package-state probes, runtime boundary loads, and import-heavy tests. It is not a broad cold-start win when every module path is unique. Unique paths still pay first-resolution cost, with cache entries capped to avoid retaining arbitrary module paths forever.

Verification

OPENCLAW_VITEST_MAX_WORKERS=1 pnpm test src/plugins/sdk-alias.test.ts src/plugins/jiti-loader-cache.test.ts
pnpm build
commit hook pnpm check

greptile-apps · 2026-04-20T12:06:08Z

Greptile Summary

This PR implements two startup-time performance optimizations: memoizing buildPluginLoaderAliasMap to avoid repeated O(N²) jiti alias normalization, and creating a narrow runtime-setter-api.ts entry point for the Slack plugin to bypass the 284 KB barrel during register(). Both changes are well-scoped and correctly implemented.

The only thing to verify is whether extensions/slack/channel-entry.ts (not touched in this PR) is also a live code path for plugin registration — if so, it still points at the wide ./runtime-api.js barrel and would miss the optimization.

Confidence Score: 5/5

Safe to merge; all remaining findings are P2 suggestions with no correctness or data-integrity impact.

The memoization logic correctly covers all inputs to the underlying resolvers, and the new Slack entry point cleanly re-exports only the needed symbol. The only open question is whether channel-entry.ts is a live path that should also be updated — that is a follow-up improvement, not a blocker.

extensions/slack/channel-entry.ts — still references the wide runtime-api.js barrel; worth confirming if this path is active.

Comments Outside Diff (1)

extensions/slack/channel-entry.ts, line 17 (link)

channel-entry.ts still references the wide barrel

channel-entry.ts is a separate entry point that still points runtime.specifier at ./runtime-api.js. If this file is used in any plugin register() path, it will continue pulling in the 284 KB / 29-chunk barrel that this PR is specifically trying to avoid. Should ./runtime-setter-api.js be used here too?

Prompt To Fix With AI

This is a comment left during a code review.
Path: extensions/slack/channel-entry.ts
Line: 17

Comment:
**`channel-entry.ts` still references the wide barrel**

`channel-entry.ts` is a separate entry point that still points `runtime.specifier` at `./runtime-api.js`. If this file is used in any plugin `register()` path, it will continue pulling in the 284 KB / 29-chunk barrel that this PR is specifically trying to avoid. Should `./runtime-setter-api.js` be used here too?



How can I resolve this? If you propose a fix, please make it concise.

Prompt To Fix All With AI

This is a comment left during a code review.
Path: extensions/slack/channel-entry.ts
Line: 17

Comment:
**`channel-entry.ts` still references the wide barrel**

`channel-entry.ts` is a separate entry point that still points `runtime.specifier` at `./runtime-api.js`. If this file is used in any plugin `register()` path, it will continue pulling in the 284 KB / 29-chunk barrel that this PR is specifically trying to avoid. Should `./runtime-setter-api.js` be used here too?

```suggestion
    specifier: "./runtime-setter-api.js",
```

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (1): Last reviewed commit: "test: add memoization safety tests for b..." | Re-trigger Greptile}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f9c19e35d1

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-04-20T13:00:50Z

+  aliasMapCache.set(cacheKey, result);
+  return result;


Cache the Jiti-normalized alias map, not the raw alias object

This memoization does not actually unlock Jiti’s normalizeAliases sentinel fast-path: Jiti tags the normalized clone with Symbol.for("pathe:normalizedAlias"), not the input alias object, so returning the same raw result reference here still causes re-normalization on each createJiti call. In practice, the startup-cost regression this commit targets remains, while we still introduce process-lifetime alias caching behavior. To make this optimization effective, the cached value needs to be the normalized alias object that already carries Jiti’s sentinel.

Useful? React with 👍 / 👎.

buildPluginLoaderAliasMap() creates a new alias object via spread on every call. jiti's normalizeAliases() uses a reference-identity sentinel (`if (e[pt]) return e`) to skip its O(N²) normalization work — but fresh object refs defeat the sentinel, causing the full cycle to repeat on every call. This change caches alias maps by their inputs (modulePath, argv1, moduleUrl, pluginSdkResolution) so identical parameters return the same object reference. Subsequent jiti calls hit the sentinel fast-path instead of re-running normalization. Includes 5 new tests covering: - reference identity for identical inputs - cache isolation (different modulePath, pluginSdkResolution, argv1 each produce distinct objects) - content equivalence between cached and freshly-computed results Refs #68983, #63948

steipete · 2026-04-20T14:17:01Z

Thanks @amknight, landed via rebase merge.

Landed on main at 8a660099f2b3c2bc4dd9a90b2820106562c97913.

openclaw-barnacle Bot added channel: slack Channel integration: slack size: XS maintainer Maintainer-authored PR labels Apr 20, 2026

amknight marked this pull request as ready for review April 20, 2026 12:04

openclaw-barnacle Bot added size: S and removed size: XS labels Apr 20, 2026

amknight changed the title ~~perf: memoize buildPluginLoaderAliasMap + narrow slack runtime setter~~ perf: memoize buildPluginLoaderAliasMap Apr 20, 2026

amknight force-pushed the perf/startup-phase1 branch from cb1d934 to 11e3b49 Compare April 20, 2026 12:15

openclaw-barnacle Bot removed the channel: slack Channel integration: slack label Apr 20, 2026

amknight mentioned this pull request Apr 20, 2026

perf(slack): narrow runtime-setter specifier to avoid loading 284KB barrel #69325

Closed

amknight changed the title ~~perf: memoize buildPluginLoaderAliasMap~~ perf: memoize buildPluginLoaderAliasMap to enable jiti sentinel reuse Apr 20, 2026

steipete changed the title ~~perf: memoize buildPluginLoaderAliasMap to enable jiti sentinel reuse~~ perf: memoize plugin loader alias maps Apr 20, 2026

chatgpt-codex-connector Bot reviewed Apr 20, 2026

View reviewed changes

openclaw-barnacle Bot added size: M and removed size: S labels Apr 20, 2026

steipete force-pushed the perf/startup-phase1 branch from dc65601 to 7fac8da Compare April 20, 2026 13:49

amknight and others added 5 commits April 20, 2026 14:50

test: cover alias map cache context

f140bd3

docs: clarify alias map memoization rationale

d413b66

perf: reuse plugin loader config cache

6cb3ce9

docs: add changelog for PR 69316

c5fc4ef

steipete force-pushed the perf/startup-phase1 branch from 7fac8da to c5fc4ef Compare April 20, 2026 13:51

steipete merged commit 8a66009 into main Apr 20, 2026
48 checks passed

steipete deleted the perf/startup-phase1 branch April 20, 2026 14:00

This was referenced Apr 21, 2026

perf(slack): narrow runtime-setter + lazy-load 4 modules + narrow 2 SDK surfaces #69317

Merged

perf(plugin-sdk): per-phase + per-jiti-call probes for bundled channel entries #69537

Merged

clawsweeper Bot mentioned this pull request Apr 27, 2026

feat(gateway): lazy-load channel plugins with opt-out config #65444

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

perf: memoize plugin loader alias maps#69316

perf: memoize plugin loader alias maps#69316
steipete merged 5 commits into
mainfrom
perf/startup-phase1

amknight commented Apr 20, 2026 •

edited by steipete

Loading

Uh oh!

greptile-apps Bot commented Apr 20, 2026 •

edited

Loading

Comments Outside Diff (1)

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Apr 20, 2026

Uh oh!

Uh oh!

steipete commented Apr 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

amknight commented Apr 20, 2026 • edited by steipete Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Why

Measurements

Tradeoff

Verification

Uh oh!

greptile-apps Bot commented Apr 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 5/5

Comments Outside Diff (1)

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Apr 20, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

steipete commented Apr 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

amknight commented Apr 20, 2026 •

edited by steipete

Loading

greptile-apps Bot commented Apr 20, 2026 •

edited

Loading