Skip to content

test(bench): document CI variance data for regression guard threshold widening (jelly-micro PR #1509) #1516

@carlos-alm

Description

@carlos-alm

Context

In PR #1509 (fix(wasm): port missing node extractions to JS extractor), tests/benchmarks/regression-guard.test.ts widened the 1-file rebuild and No-op rebuild thresholds from 25%→50% and CANARY mode to 100%.

The justification is empirical CI variance on sub-30ms metrics (±10ms CI jitter ≈ ±50%+ relative), and MIN_ABSOLUTE_DELTA provides a floor. However, the PR does not include CI data showing the measured variance.

Action

  • Attach a CI run showing the variance on 1-file rebuild, No-op rebuild, and CANARY metrics that motivated the threshold change.
  • If the empirical data supports the change, add a comment in the test file citing the CI run ID.
  • If the data does not support it, revert to 25% and investigate the root cause of the instability.

Metadata

Metadata

Assignees

No one assigned

    Labels

    ciContinuous integration / buildfollow-upDeferred work from PR reviews that needs tracking

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions