[lstein]

Summary

frontend-webv2-tests fails on main since 501071e (#366), and on every PR that has merged main since (#367, twice), on one browser test:

WorkflowLinearPanel.browser.test.tsx › Linear form field entry › keeps the caret while editing an exposed text field through the project graph

The CI log shows the Prompt input was ready ("element is visible, enabled and stable"). But a chakra-tooltip__positioner from the form itself "intercepts pointer events" on every retry of locator.click, until the 15 s timeout. Each exposed field's label and reset button carries a tooltip, and on the runner one opened over the input.

The test is about the caret surviving an edit's round trip through the project graph. It clicked each input only to focus it. Its focusAtEnd helper now focuses the input directly and presses End, so there is no hit-test for a tooltip to intercept. The three tests using the helper still type, move the caret and assert selection exactly as before.

Related Issues / Discussions

Unblocks #367, and any PR that merges main.

QA Instructions

• pnpm -C invokeai/frontend/webv2 test:browser src/features/workflow/ui/linear/WorkflowLinearPanel.browser.test.tsx: 9 passed, 3 of 3 runs.

• lint:oxc and lint:tsc: clean.

• The failure does not reproduce locally. Before the change the test passed in every setup tried:

  • the test alone;
  • together with fix(webv2): refresh saved workflow picker after library changes #366's new CallSavedWorkflowSyncRuntime test;
  • the full browser suite pinned to 4 CPUs;
  • the pointer forced to the viewport origin;
  • the Prompt label hovered for 1.5 s before the click.

  The evidence for the cause is the CI call log. This PR's own CI run is the confirmation.

Review

Considered and rejected: a global beforeEach that parks the pointer in a corner before every browser test. It would add a hover to ~1800 tests, and could not be shown to fix this case. The change here stays in the one test helper whose click was incidental.

Checklist

• The PR has a short but descriptive title, suitable for a changelog
• Meaningful regression coverage added / updated where needed; obsolete tests/code removed
• Persisted-state and API changes include required migrations / compatibility validation
• Relevant performance/efficiency opportunities considered; material claims have evidence
• Material review findings resolved and relevant checks rerun
• Documentation added / updated (if applicable)
• Updated What's New copy (if doing a release after this PR)