Repository navigation
test(e2e): keep two host sections in the catalog-retry reconnect test - #26838
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configuration
📒 Files selected for processing (1)
Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 9 remain after this review. 📝 WalkthroughWalkthroughThe catalog-failure reconnection test now requests the Priority: ⬇️ Low Merge Risk: ⚪ Minimal · up to This test-only change supplies the second host context while retaining the catalog-failure and reconnect checks. No actionable merge risk remains. 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)✨ Finishing Touches📝 Generate docstrings
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 |
ELI5
One of the SSH end-to-end tests on main started failing on every pull request. The test was waiting for a sidebar heading that the app correctly stops showing. This PR gives the test the setup it needs to see that heading again, and leaves the app unchanged.
What Changed
The test "reconnect restores the managed host name after its conversion catalog fails" checks that, after the managed-server list briefly fails to load and then recovers, the sidebar heading for the converted SSH host shows the host's real name.
The sidebar only shows host headings when workspaces are split across at least two hosts. A single host section is hidden on purpose. In this test, both seeded workspaces (a project and a folder) live on the one SSH host.
Why
The test came from #26531, which branched before #26535 landed. On that older base, a migrated folder was wrongly kept under its old SSH host, so the folder and the project split into two host sections by accident, and that is the only reason the heading appeared. #26535 fixed that bug, and its sibling test (
a relay-era profile converts its SSH host...) added a local project at the same point for the same reason. This PR uses the same setup. It adds no waits, and the assertions stay as strict as before: the folder row must keep its managed host identity, and the heading must show the recovered host name.The other option was to change the product so a single host section still shows its heading. That would bring back the clutter the "lone section is hidden" rule exists to prevent, so I rejected it.
Linked Issue
N/A (maintainer fix for main CI)
Visual Proof
N/A: test-only change. CI failure screenshot from main showed the project and the folder grouped together with no host heading.
Testing
ssh host auto-converts to managed orcad (Docker)runs this spec)check:code-quality:changed,audit:anti-slopandpnpm tcpass locallyReview
Agent skill upstream boundary
Notes
Test-only. No impact on platforms, SSH behavior, or compatibility.
Checklist
N/Awith reasonpnpm lint,pnpm typecheck,pnpm test, andpnpm buildpass (or CI will cover; local preferred)