Skip to content

e2e: the Listing reopen test waits for its run before it ends - #45

Merged
wilsonhj merged 1 commit into
mainfrom
fix/e2e-listing-reopen-leak
Oct 30, 2026
Merged

wilsonhj merged 1 commit into
mainfrom
fix/e2e-listing-reopen-leak

Conversation

@wilsonhj

Copy link
Copy Markdown
Owner

Problem

On main, the full e2e suite fails one test: motion.spec.ts:37 (person + product -> one Final job) expects 1 submitted graph and gets 4.

The extra three are Listing panel jobs. The test added in #44 (listing.spec.ts, "reopening a run with ?run= ...") starts a Story · 8 run and ends while the server is still submitting panel jobs. motion.spec runs next and resets the mock in beforeEach, and the late Listing submissions then land in its request log. CI runs no e2e, so CI stayed green.

Repro: pnpm exec playwright test tests/e2e/listing.spec.ts:231 tests/e2e/motion.spec.ts:37 fails every time on main.

Fix

The reopen test now waits for all 8 tiles (expectAllPanelsDone), as the other Listing tests do, before it opens the deep link.

Verification

  • The two-test repro: 2 passed (failed before).
  • Full pnpm exec playwright test: 39 passed, 8 skipped, 0 failed (1 failed before, in 2 of 2 runs).
  • pnpm lint and pnpm typecheck clean.

🤖 Generated with Claude Code

The test added in #44 started a Story · 8 run and finished while panel
jobs were still being submitted. Those late submissions landed in the
mock's request log after motion.spec reset it, so "person + product ->
one Final job" saw 4 graphs instead of 1 whenever the full suite ran
(CI runs no e2e, so it stayed green).

Wait for all 8 tiles, as the other Listing tests do, before reopening.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@wilsonhj
wilsonhj merged commit 8cb80d5 into main Oct 30, 2026
1 check passed
wilsonhj added a commit that referenced this pull request Oct 31, 2026
* CI: run the Playwright e2e suite against the mock Comfy API

Its own job, so a cross-spec leak like the one PR #45 fixes fails a PR
instead of only full local runs. Uploads test-results/ and the HTML
report when it fails.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* TEMP: break a smoke assertion to prove CI e2e fails (will be reverted)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Revert the temporary broken smoke assertion

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
wilsonhj added a commit that referenced this pull request Oct 31, 2026
The test added in #44 started a Story · 8 run and finished while panel
jobs were still being submitted. Those late submissions landed in the
mock's request log after motion.spec reset it, so "person + product ->
one Final job" saw 4 graphs instead of 1 whenever the full suite ran
(CI runs no e2e, so it stayed green).

Wait for all 8 tiles, as the other Listing tests do, before reopening.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
wilsonhj added a commit that referenced this pull request Oct 31, 2026
Its own job, so a cross-spec leak like the one PR #45 fixes fails a PR
instead of only full local runs. Uploads test-results/ and the HTML
report when it fails.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
wilsonhj added a commit that referenced this pull request Oct 31, 2026
* CI: run the Playwright e2e suite against the mock Comfy API

Its own job, so a cross-spec leak like the one PR #45 fixes fails a PR
instead of only full local runs. Uploads test-results/ and the HTML
report when it fails.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* TEMP: break a smoke assertion to prove CI e2e fails (will be reverted)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Revert the temporary broken smoke assertion

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant