Uh oh!
There was an error while loading. Please reload this page.
test: TKT-305 CI control probe (no-op) - #1
Closed
sepo-eng wants to merge 2 commits into
Closed
Conversation
This PR doesn't fully meet our contributing guidelines and PR template. What needs to be fixed:
Please edit this PR description to address the above within 2 hours, or it will be automatically closed. If you believe this was flagged incorrectly, please let a maintainer know. |
Thanks for your contribution! This PR doesn't have a linked issue. All PRs must reference an existing issue. Please:
See CONTRIBUTING.md for details. |
6 tasks
sepo-eng
commented
Aug 10, 2026
Author
Control experiment for TKT-305 complete — evidence captured (blacksmith runners confirmed stuck queued, GitHub-hosted confirmed working). Closing this throwaway probe. |
This was referenced Aug 12, 2026
sepo-eng added a commit
that referenced
this pull request
Aug 13, 2026
…T-411) (#49) * fix(app): guard shell settings against a non-array resource value (TKT-411) Sean's live gate walk / feedback anomalyco#202: opening the settings-v2 dialog could crash the whole app (app-root ErrorBoundary swallows the window) with "input.shells.reduce is not a function". Root-caused to two independent defects, both fixed: 1. e2e/utils/mock-server.ts modeled `/api/pty/shells`, but the app calls the real bare V1 endpoint `/pty/shells` (confirmed: no `/api/pty/shells` exists in packages/sdk/openapi.json). Every request fell through to the fixture's generic fallback, returning a bare `{}` -- not caught by `?? []`. 100% deterministic, not timing-dependent: a normal (wait for app-visible, then open settings) flow crashed identically. No existing spec caught it because none opens the General tab under v1 protocol. Also fixed the response body shape while at it -- the real endpoint returns a bare array, not the `{location, data}` envelope most other endpoints in this fixture use. 2. A genuine, separate timing defect (the ticket's original theory, confirmed real independent of #1 via delete-the-fix): opening the dialog before the app finishes its own startup still crashes even with a correct array response. Guarded at the source in general-controllers.ts, widened from a bare `?? []` to `Array.isArray(...) ? ... : []` -- #1 proved "guard only undefined" isn't sufficient against every wrong shape a response can take. New regression spec asserts both are fixed together; delete-the-fix verified against a 3-configuration matrix (each fix independently necessary, neither alone sufficient). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M1CJmqai46ihahdKMVNB27 * docs(app): correct the overstated race claim on TKT-411's regression spec Henry's review: re-running my delete-the-fix matrix with a genuinely fresh dev server per configuration (mine had reuseExistingServer contamination between arms) shows the committed test discriminates only the mock-server.ts fixture bug -- fixing that alone is sufficient to pass, with or without the general-controllers.ts guard. My original comment's "confirmed by delete- the-fix" claim for an independent timing race was not actually demonstrated by this test. Softened to state exactly what's proven (the fixture bug, deterministically) versus what's reasoned-but-undemonstrated (the controller guard, kept as a defensible "prefer unrepresentable" improvement per the ticket's own fix shape, not backed by a red/green in this file). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M1CJmqai46ihahdKMVNB27 * fix(app): wait for app-visible before opening settings in TKT-411's regression spec Henry's review: the spec's immediate Control+, fire (a leftover from the retracted timing-race hypothesis) is itself flaky on a cold server -- CI's own configuration, reuseExistingServer: false -- failing ~2/3 runs and only appearing green in CI because retries:2 absorbed it. Root cause: firing the shortcut before the app finishes booting races app startup itself, independent of anything this PR is about. Fixed to expectAppVisible before the keypress, matching the established pattern (settings-skills-catalog.spec.ts, remote-session-settings.spec.ts). Reverified deterministic in both directions with a genuinely fresh dev server (port-listener checked, not assumed) per run: 4/4 pass on fixed code, 3/3 fail on pristine code, no retries needed either way. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M1CJmqai46ihahdKMVNB27 --------- Co-authored-by: opencode <bot@opencode.ai> Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Control experiment for TKT-305 (CI/review baseline on fork PRs). No functional change — appends a comment to CONTEXT.md. Verifying which upstream workflows actually fire and produce check-runs on a PR against this fork. Will close/delete after evidence is captured.