Skip to content

v0.2.0 — Import and worker comparisons

Latest

Choose a tag to compare

@Issey-Kubota Issey-Kubota released this 11 Oct 06:56
8d16612

What is new

  • Compare Vitest worker limits with npm run measure:workers: limits of 1, 2, and 4 are interleaved over 15 runs, with five samples per condition.
  • Record median, minimum, and maximum wall times for each condition while keeping imports and test inputs fixed. CPU timing remains optional; Windows uses wall time only.
  • A unified English README presents both experiments, setup, output interpretation, limitations, and development instructions.
  • The import comparison remains available through npm run measure (six alternating runs).

Try the worker comparison

Download and extract the source archive below, open a terminal in its directory, and run:

npm ci
npm run verify:manifest
npm run measure:workers
npm run verify:manifest

Run the measurement command once. It executes all 15 measurements internally. Use a dedicated working copy with no concurrent edits. maxWorkers is an upper limit, not a guarantee that every worker is active.

See the README and worker comparison guide for requirements and interpretation. Existing import-only articles can continue to use v0.1.0.

Validation scope

The development checks passed 36 helper regression tests, including the 15-run worker schedule and unchanged test inputs. The maintainer reported a successful 15-run Windows measurement with no invalid runs. Its measurement commit and detailed report have not yet been independently confirmed; those reported timings are not a benchmark of this release. No new performance run or CI run was performed for this release. The existing CI workflow covers the import comparison, not the worker comparison. macOS remains unverified on real hardware.

Source commit: 8d166124870ddf9f075a04de73af6ffc32bbbf93.