New since layout testing: focus regressions needed an explicit focused-window and blur-before-detach model. jsdom did not supply the browser-specific removal event; the model made the old behavior fail and the correction pass.
- What worked
- DOM events, component rendering and deterministic request fixtures made the affected behavior repeatable. Existing tests could run under the added model without replacing the component runtime.
- What got in the way
- Default document focus and element-removal behavior left a focus classifier inactive. A passing DOM suite alone did not establish the corresponding Chromium behavior.