bench(startup): add ESM variant that exercises the iitm loader#8956
Merged
Conversation
The startup variants load the fixture through CommonJS require, which goes through require-in-the-middle and never registers the iitm ESM loader. That loader is what the synchronous module.registerHooks work changes from an off-thread async loader to an in-thread sync one, so the suite reported no movement for that change: nothing put the loader on a measured path. with-tracer-everything-esm registers the loader via --import ../../../register.js and imports the same fixture through ESM, so every dependency and its transitive graph flow through the loader's resolve/load hooks. The fixture's index.mjs reads the same dependencies as index.js, so the dependency list stays single-sourced in package.json. Refs: #8942
Contributor
Overall package sizeSelf size: 6.25 MB Dependency sizes| name | version | self size | total size | |------|---------|-----------|------------| | import-in-the-middle | 3.0.2 | 85.93 kB | 825.11 kB | | opentracing | 0.14.7 | 194.81 kB | 194.81 kB | | dc-polyfill | 0.1.11 | 25.74 kB | 25.74 kB |🤖 This report was automatically generated by heaviest-objects-in-the-universe |
BenchmarksBenchmark execution time: 2026-06-17 13:45:19 Comparing candidate commit 4766a8f in PR branch Found 0 performance improvements and 0 performance regressions! Performance is the same for 1950 metrics, 15 unstable metrics.
|
BridgeAR
marked this pull request as ready for review
June 17, 2026 13:48
juan-fernandez
approved these changes
Jun 17, 2026
dd-octo-sts Bot
pushed a commit
that referenced
this pull request
Jun 18, 2026
The startup variants load the fixture through CommonJS require, which goes through require-in-the-middle and never registers the iitm ESM loader. That loader is what the synchronous module.registerHooks work changes from an off-thread async loader to an in-thread sync one, so the suite reported no movement for that change: nothing put the loader on a measured path. with-tracer-everything-esm registers the loader via --import ../../../register.js and imports the same fixture through ESM, so every dependency and its transitive graph flow through the loader's resolve/load hooks. The fixture's index.mjs reads the same dependencies as index.js, so the dependency list stays single-sourced in package.json. Refs: #8942
Closed
tlhunter
pushed a commit
that referenced
this pull request
Jun 18, 2026
The startup variants load the fixture through CommonJS require, which goes through require-in-the-middle and never registers the iitm ESM loader. That loader is what the synchronous module.registerHooks work changes from an off-thread async loader to an in-thread sync one, so the suite reported no movement for that change: nothing put the loader on a measured path. with-tracer-everything-esm registers the loader via --import ../../../register.js and imports the same fixture through ESM, so every dependency and its transitive graph flow through the loader's resolve/load hooks. The fixture's index.mjs reads the same dependencies as index.js, so the dependency list stays single-sourced in package.json. Refs: #8942
Merged
pabloerhard
pushed a commit
that referenced
this pull request
Jun 26, 2026
The startup variants load the fixture through CommonJS require, which goes through require-in-the-middle and never registers the iitm ESM loader. That loader is what the synchronous module.registerHooks work changes from an off-thread async loader to an in-thread sync one, so the suite reported no movement for that change: nothing put the loader on a measured path. with-tracer-everything-esm registers the loader via --import ../../../register.js and imports the same fixture through ESM, so every dependency and its transitive graph flow through the loader's resolve/load hooks. The fixture's index.mjs reads the same dependencies as index.js, so the dependency list stays single-sourced in package.json. Refs: #8942
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The
startupsirun variants loadeverything-fixturethrough CommonJSrequire, which goes through require-in-the-middle and never registers the iitm ESM loader. That loader is what #8942 changes from an off-thread asynchronous loader (module.register) to an in-thread synchronous one (module.registerHooks), so the benchmark suite reported "performance is the same" for that PR — no variant put the loader on a measured path.with-tracer-everything-esmregisters the loader via--import ../../../register.jsand imports the same fixture through ESM, so every dependency and its transitive graph flow through the loader's resolve/load hooks. On master this measures the async loader; on #8942 (sync loader on capable Node) the A/B renders the difference.index.mjsreads the samedependenciesasindex.js, so the dependency list stays single-sourced inpackage.json— updating one updates both.registerHooks, so perf(loader): use synchronous iitm hooks when available #8942 keeps the async loader there (no delta); 24/26 show the sync win.Why
#8942's 56% Vitest startup reduction was invisible to CI because the existing startup variants are CJS-only. This closes that coverage gap and guards the ESM loader path against future regressions.
Test plan
startup / with-tracer-everything-esmreports stableinstructionsandwall.timeon Node 20/24/26.Note on local validation: macOS sirun hangs on every
startupvariant (including the existing ones), so stability was checked with direct process timing — steady-state ~0.69s, ~2% spread, all three variants exit 0 on Node 20/24/26 — and the sirun numbers are left to CI's pinned isolated core, per the benchmark guidance.Refs: #8942