Separate command-bar setting and shortcut coverage

[JordanNoone]

What this changes

Reopen the command palette through its in-app Commands button in the multi-step setting test. Keep real Mod+K coverage in the dedicated keybinding test immediately below it.

In plain English: Ubuntu desktop CI runs four Electron windows on one shared display. The setting test does not need a second focus-sensitive shortcut assertion before it can test debug-panel arguments; a neighboring test already owns that behavior and explicitly focuses the editor first.

Fixes #13468.

Why

The affected setting test failed on its first execution in all seven sampled hourly Ubuntu runs, then passed under the isolated --last-failed retry.

Temporary capture-phase logging established the boundary:

• Four-worker first run: no Mod+K event reached the app keymap listener, and the command machine remained Closed.
• One-worker isolated retry: the keymap resolved command-palette.open and the machine entered Selecting command.

Diagnostic evidence: https://github.com/KittyCAD/modeling-app/actions/runs/33320150808/job/99281435652

This disproved stale keymap-scope and delayed transition-cleanup hypotheses. page.bringToFront() also proved insufficient: it passed once, then the clean main rerun lost focus again before the key event.

Bring-to-front counterexample: https://github.com/KittyCAD/modeling-app/actions/runs/33321342642/job/99283913271

All speculative product changes and temporary instrumentation are removed.

Test goal / intent

• Setting-flow test: prove the command palette can change the debug-panel setting and move between its arguments.
• Dedicated keybinding test: prove Mod+K opens and focuses the palette from the code editor.

The tests now express those goals independently instead of making the setting flow depend on a second shared-window shortcut assertion.

Validation

• Biome formatting, ESLint, and git diff --check pass for the final changed test.
• Existing diagnostics prove the failure is before app keymap dispatch, not inside the command machine.
• Two independent four-worker Ubuntu shard 1 runs passed both relevant tests in their initial batches. Each ended with 41 passed and no --last-failed phase.
• PR check run: https://github.com/KittyCAD/modeling-app/actions/runs/33321728739/job/99284875258
• Independent workflow-dispatch run: https://github.com/KittyCAD/modeling-app/actions/runs/33322038606/job/99286324085
• The PR's full E2E workflow passed across Ubuntu, macOS, Windows, web, desktop, and snapshots.
• The branch is one commit directly on merged main, with one changed test file.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
modeling-app	Ready	Preview	Aug 31, 2026 2:09am

[JordanNoone]

First-run desktop validation passed. Plain-English goal: ensure the Electron window owned by this test is active before verifying its window-level Mod+K shortcut. The exact scenario passed in the initial four-worker Ubuntu batch in 5.4 seconds; the shard retry ran only an unrelated -XY sketch test. Evidence: https://github.com/KittyCAD/modeling-app/actions/runs/33320795024/job/99282979477

[JordanNoone]

Final validation supersedes the earlier bringToFront() comment: foreground activation was not atomic across parallel Electron workers, so that approach was removed. The final one-file diff separates setting-flow and shortcut responsibilities. Two independent Ubuntu shard 1 runs passed both tests on their first execution and ended with all 41 tests passing, without --last-failed: https://github.com/KittyCAD/modeling-app/actions/runs/33321728739/job/99284875258 and https://github.com/KittyCAD/modeling-app/actions/runs/33322038606/job/99286324085.