Problem
The full post-merge E2E run shows BigData row-pinning / thumb-drag scroll coverage failing while the analogous DevIndex scroll tests pass. This needs a focused fix or contract correction so scroll pinning regressions do not stay hidden behind a broad red suite.
Evidence
Verified on dev at merge commit ecbb21cde9a72f40f7a6c7132baedecd25db6bdd after PR #14849 merged.
Command run:
./node_modules/.bin/playwright test -c tmp/e2e-full-8097.config.mjs
The temporary config matched the E2E config but used http://localhost:8097 with reuseExistingServer: false, because port 8080 was already occupied by another Neo checkout.
Suite result: 73 tests run, 56 passed, 17 failed.
Failure family:
test/playwright/e2e/grid/RowPinning.spec.mjs:278: expected blankFrames to be 0, received 185 in Scroll Telemetry: Visual Blanking and Jitter Detector.
test/playwright/e2e/grid/ThumbDragPause.spec.mjs:45: expected window.__BLANK to be true after pausing during thumb drag, received false.
The same run passed test/playwright/e2e/grid/ThumbDragDevIndex.spec.mjs vertical blanking and horizontal virtualization coverage, so the failure appears scoped to the BigData row-pinning / pause path rather than every scroll telemetry spec.
Acceptance Criteria
- BigData row-pinning scroll telemetry either reports zero blank frames again or the test is deliberately updated to the current supported invariant with a documented reason.
- The thumb-drag pause behavior is made consistent with the intended pinning timeout contract and
ThumbDragPause.spec.mjs asserts that contract.
- Focused rerun passes:
./node_modules/.bin/playwright test -c test/playwright/playwright.config.e2e.mjs test/playwright/e2e/grid/RowPinning.spec.mjs test/playwright/e2e/grid/ThumbDragPause.spec.mjs
Duplicate Sweep
- Latest 30 open issues checked: no matching open ticket.
- Targeted GitHub search checked:
RowPinning ThumbDragPause blankFrames pinning returned no open issues.
- Local synced issue/discussion cache checked for
RowPinning and ThumbDragPause; hits were historical/archived prior work, not an open active duplicate.
Problem
The full post-merge E2E run shows BigData row-pinning / thumb-drag scroll coverage failing while the analogous DevIndex scroll tests pass. This needs a focused fix or contract correction so scroll pinning regressions do not stay hidden behind a broad red suite.
Evidence
Verified on
devat merge commitecbb21cde9a72f40f7a6c7132baedecd25db6bddafter PR #14849 merged.Command run:
./node_modules/.bin/playwright test -c tmp/e2e-full-8097.config.mjsThe temporary config matched the E2E config but used
http://localhost:8097withreuseExistingServer: false, because port 8080 was already occupied by another Neo checkout.Suite result: 73 tests run, 56 passed, 17 failed.
Failure family:
test/playwright/e2e/grid/RowPinning.spec.mjs:278: expectedblankFramesto be0, received185inScroll Telemetry: Visual Blanking and Jitter Detector.test/playwright/e2e/grid/ThumbDragPause.spec.mjs:45: expectedwindow.__BLANKto betrueafter pausing during thumb drag, receivedfalse.The same run passed
test/playwright/e2e/grid/ThumbDragDevIndex.spec.mjsvertical blanking and horizontal virtualization coverage, so the failure appears scoped to the BigData row-pinning / pause path rather than every scroll telemetry spec.Acceptance Criteria
ThumbDragPause.spec.mjsasserts that contract../node_modules/.bin/playwright test -c test/playwright/playwright.config.e2e.mjs test/playwright/e2e/grid/RowPinning.spec.mjs test/playwright/e2e/grid/ThumbDragPause.spec.mjsDuplicate Sweep
RowPinning ThumbDragPause blankFrames pinningreturned no open issues.RowPinningandThumbDragPause; hits were historical/archived prior work, not an open active duplicate.