Current Behavior
Failures currently contribute to the flaky count even when the PR modified the failing test. Such failures are likely caused by the PR and should not count as independent flaky occurrences.
For example, CI build 75799 failed test-ffi-fast-buffer, but the PR modified test/ffi/test-ffi-fast-buffer.js. Excluding this occurrence would leave only one qualifying PR, so the test would not appear in the reliability report:
https://github.com/nodejs/reliability/blob/main/reports/2026-08-15.md
Expected Behavior
Exclude an occurrence from the flaky count when its test file was modified by the tested PR revision.
Current Behavior
Failures currently contribute to the flaky count even when the PR modified the failing test. Such failures are likely caused by the PR and should not count as independent flaky occurrences.
For example, CI build 75799 failed
test-ffi-fast-buffer, but the PR modifiedtest/ffi/test-ffi-fast-buffer.js. Excluding this occurrence would leave only one qualifying PR, so the test would not appear in the reliability report:https://github.com/nodejs/reliability/blob/main/reports/2026-08-15.md
Expected Behavior
Exclude an occurrence from the flaky count when its test file was modified by the tested PR revision.