Repository navigation
test: share floating panel behavior suite startup - #29304
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info
📝 Walkthrough
Priority: ⬇️ Low Merge Risk: ⚪ Minimal · up to
|
| Check name | Status | Explanation |
|---|---|---|
| Docstring Coverage |  |
Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 1 functions across 3 files. |
| Linked Issues check | |
Check skipped because no linked issues were found for this pull request. |
| Out of Scope Changes check | |
Check skipped because no linked issues were found for this pull request. |
| Title check | |
The title clearly and concisely describes the main change: sharing startup work for the floating-panel behavior test suite. |
| Description check | |
The description follows the required template and explains the change, rationale, testing, visual impact, review notes, and checklist status. It provides detailed validation results and clearly states… |
Full details: Description check
-
Explanation
The description follows the required template and explains the change, rationale, testing, visual impact, review notes, and checklist status. It provides detailed validation results and clearly states the remaining limitations and CI evidence.
✨ Finishing Touches
-
📝 Generate docstrings
-
- Commit to this branch
- Create a new PR
-
- Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.
Comment @coderabbitai help to get the list of available commands.
ELI5
Floating-panel tests repeatedly load the same component and register the same mocks. Run their 73 cases together so the PR suite does that work once.
What Changed
Four behavior suites become imported specs behind one test entry point. The assertions stay with their existing behavior areas; 35 common mock declarations register once. Setup resets the markdown creator and gives each case a fresh focus IPC function, using the production cache's existing identity boundary. Remove the cache-reset function and re-export used only by tests.
Why
The panel behaves the same, with the same bounds, focus, shortcuts, and tab-lifecycle regressions protected. Sharing setup avoids four isolated import graphs without reducing assertions or adding a production testing hook. Test/support drops 531 lines; production drops six.
Linked Issue
Maintainer-requested CI test audit. #29298 has landed. This PR now includes current main and targets main.
Visual Proof
N/A — test setup consolidation and removal of an unused product export; no rendered behavior changes.
Testing
Fresh verification after merging main 9c177fb: all 108 owner/sibling cases pass, including all 73 panel cases. Actual dynamic collection and all six measurement reports preserve the exact original 73 full names, once each. Shuffled panel suite passes all 73 (seed 104729). Web typecheck, all changed-code-quality checks, formatting and whitespace checks pass. No native-focus/visible-window tests were run.
Three alternating full-file pairs on this reconciled source, Linux, Bun 1.4.3 / Node 24.21.0, one worker, module cache off and background launch:
Every run passes the same 73 cases; baseline restores only this PR's paths from current main while the candidate uses the consolidated entry. This is a controlled local observation, not a whole-PR wall claim. Historical hosted selected module time was 4,944 → 981 ms (runs 38827086842 / 38828680027; verified actual sources), with all five reports clean and matching Node/Bun versions/four workers on ARM. That unpaired older comparison has runner/base/cache/workload confounders and does not validate the reconciled head. Fresh reconciled-head full PR run 38855578573 passed all five unit shards with zero unhandled errors across 12,216 executed files. The actual tested merge tree matches the reviewed head; its wrapper passed all 73 cases, with 568 ms official module cost versus 5,202 ms summed across the four original suites in run 38848025584. This historical, unpaired comparison changed from GitHub ARM / Node 24.21 to Blacksmith X64 / Node 24.19; base code, dependencies, workload, caches and scheduling also differ. It does not isolate this refactor or establish whole-PR savings. Full required CI remains pending before merge.
Review
All original cases retained. Shared browser-close mock preserves the lifecycle suite's behavior; other suites do not exercise a nonempty browser close. Per-case reset prevents shared mock implementation leakage. Independent preservation review passed: no missing contracts, changed mock behavior, or discovery gaps.
Agent skill upstream boundary
Checklist
Refresh validation: head 1ed0341 includes current main 98457f6 and merged Kitty fixture repair #29519. The complete floating subtree, dependency lock, package and Vitest configuration are unchanged from independently approved bb2cde0. Eight affected owner/drag/layout/producer suites pass 127 cases; renderer typecheck, changed-code quality and whitespace checks pass. The prior dynamic collection/shuffle proof still covers identical floating/runner bytes.
Previous head bb2cde0 full PR run 38868611124 passed aggregate 116869988453 and all five clean unit shards; actual-source qualification confirms all 73 wrapper cases passed, with official module duration 797 ms. That run skipped real WSL, so it did not resolve the earlier image failure. Separate #29519 subsequently passed all 12 Windows/WSL/GPU image cases plus nine WSL participation cases, with retained screenshots reviewed, before merging. Fresh full CI and explicit Windows/WSL image validation at this refreshed head remain required. Prior timing comparisons stay historical; no new causal speed claim.
Final exact-head qualification: full PR workflow 38870300944 (
.github/workflows/pr.yml) passed aggregateverify/ PR checks passed 116875467802 for 1ed0341. All five clean unit reports contain 12,321 unique files exactly once; all 73 panel cases passed (424 ms official module time, 273 ms raw execution, separate metrics). The tested merge and reviewed head have identical trees. Separate exact-SHA Windows run 38870298789 passed 12 native/WSL/GPU image cases and nine WSL participation cases with zero skips or retries. All 36 screenshots were independently reviewed. This supersedes the pending status above; the earlier image failure was not waived. Current measurements verify execution and do not establish causal hosted or whole-PR savings.