Infrastructure: Restore Windows Storybook focus-regression parity #4347

Goal

Restore the complete regression coverage from the retired Windows Jest Storybook harness through the supported native Storybook/WDIO path. Parent: #4275. Follow-up to the completed retirement milestone #4254.

Status - 2026-10-01

Open; retirement is complete, regression parity is not. PR #4294 removed the legacy harness. PRs #4306, #4313, #4314, and #4320 delivered native providers, executable story tests, modern focus behavior, and FocusZone regressions.

The current catalog has 29 named WDIO cases across 17 stories, including Button, Switch, Input, TabList, and FocusZone focus behavior. Those newer tests are valuable but do not recreate the original ten delayed focus-survival cases. The current story callbacks/helpers contain no original three-second post-activation survival cohort.

Required historical coverage

The load-bearing regression from PR #4208 was a delayed 0xc0000409 application termination after focus acquisition. Preserve a minimum three-second post-activation observation, native focus confirmation, and terminal app-crash detection for each original consumer:

Also reconcile the original Button/Icon/Callout visibility and Callout native-window status checks with the replacement coverage. Use generated current story IDs and stable native testIDs, not the obsolete IDs/XPath selectors from the removed harness.

Acceptance criteria

• Each of the ten delayed cases is authored as a named, typed, component-colocated WDIO case and retains the three-second regression window.
• The Button real-Tab movement case and original visibility/status assertions have active replacement coverage.
• The tests verify native focus and application survival, not only JavaScript state or immediate renderability. Crashes during the observation window terminate the run as failures.
• Real pointer activation is qualified on an unlocked interactive Windows Fabric endpoint. Programmatic focus can supplement, but must not be mislabeled as a physical-click regression pass.
• All required cases execute in an active Windows Fabric CI lane, unquarantined, with actual pass evidence. Hosted keyboard/pointer skips do not establish parity.
• Accordion's currently excluded Fabric module has a safe, documented native test path; exclusions are not silently counted as coverage.
• Changes to FocusVisual or its consumers trigger/review this cohort, and run artifacts identify each executed/skipped case and exact app/helper/manifest identity.
• Ownership-safe teardown, process/input recovery, documentation, and relevant package checks remain intact. Do not restore Appium or the retired Jest harness.

Evidence and coordination

• Original regression: PR #4208, commit 4777e54fee2eab72af36d3bfac58e618ed0ad634.
• Current Button tests
• Shared native focus helpers
• Current authored-test contract
• Coordinate authoritative input/CI qualification with Infrastructure: Qualify and require native Storybook CI coverage #4349, native catalog exclusions with Infrastructure: Restore Storybook stories excluded by native desktop crashes #4344, and accessibility/native-state coverage with Infrastructure: Complete portable Storybook accessibility coverage and reporting #4343/Infrastructure: Fix native accessibility-state projection gaps on macOS and Win32 #4346.