e2e/poll-flow.test.js fully red: .fab never appears — no green browser-level poll path exists (0/2 since before the cordn fork) #86

New issue

Open

opened 2026-08-04 12:34:42 +00:00 by laoc · 0 comments

laoc commented

2026-08-04 12:34:42 +00:00

Owner

Symptom

Both tests in e2e/poll-flow.test.js (Poll FAB - Smoke: button is wired into the FAB, Poll - Full happy path: poll -> vote -> tally updates) fail identically: expect(locator('.fab')).toBeVisible() times out after 15s — the FAB never appears in the e2e environment.

Attribution (measured)

Fork base 1f61b22d: 0 passed / 2 failed (TestOER baseline 2026-08-04).
Branch tip b924a4a6 (feat/concord-render-gaps): same 2 failures, spec byte-identical.
src/lib/components/shared/GlobalFAB.svelte exists in both trees, so it is not a missing-component/stale-base story.

So poll e2e coverage has been fully red since before the cordn stack forked — nobody currently has a passing end-to-end poll test. That matters more now that Concord poll rendering + the PollMessage re-seed fix shipped on feat/concord-render-gaps: unit/component coverage there is good (mutation-checked), but there is no green browser-level poll path on either side of the fork.

(Discussion in Buzz #edufeed-app thread 734e3298.)

Open question / next step

Why does .fab never render in the e2e run — auth/route gating in the sandbox env, or selector drift after the #39 FAB retargeting (2d94d16b)? Whoever picks this up: reproduce with npx playwright test e2e/poll-flow.test.js at base, then check what conditions gate GlobalFAB's render on the route the spec visits.

## Symptom Both tests in `e2e/poll-flow.test.js` (`Poll FAB - Smoke: button is wired into the FAB`, `Poll - Full happy path: poll -> vote -> tally updates`) fail identically: `expect(locator('.fab')).toBeVisible()` times out after 15s — the FAB never appears in the e2e environment. ## Attribution (measured) - Fork base **1f61b22d**: **0 passed / 2 failed** (TestOER baseline 2026-08-04). - Branch tip b924a4a6 (feat/concord-render-gaps): same 2 failures, spec byte-identical. - `src/lib/components/shared/GlobalFAB.svelte` **exists** in both trees, so it is not a missing-component/stale-base story. So poll e2e coverage has been fully red since before the cordn stack forked — nobody currently has a passing end-to-end poll test. That matters more now that Concord poll rendering + the PollMessage re-seed fix shipped on feat/concord-render-gaps: unit/component coverage there is good (mutation-checked), but there is **no green browser-level poll path on either side of the fork**. (Discussion in Buzz #edufeed-app thread 734e3298.) ## Open question / next step Why does `.fab` never render in the e2e run — auth/route gating in the sandbox env, or selector drift after the #39 FAB retargeting (`2d94d16b`)? Whoever picks this up: reproduce with `npx playwright test e2e/poll-flow.test.js` at base, then check what conditions gate GlobalFAB's render on the route the spec visits.

No labels

bug

discussion

enhancement

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

edufeed/edufeed-app#86

No description provided.

Rows
Columns