FE-1578: Replace fixed sleeps in Playwright tests with condition-based waits

[claude]

Requested by Tim Diekmann · Slack thread

🌟 What is the purpose of this PR?

Before: four Playwright specs under tests/hash-playwright paused with fixed sleep(...) calls — 100 ms between editor keystrokes, 500 ms before clicking buttons, 2 s before a reload, 12 s for a notification poll. On a loaded runner the app can take longer than the sleep, so the next step runs too early and the test fails; in the merge queue that dequeues the whole batch. When the app is faster, the sleep only adds wall time.

After: each sleep is replaced by a wait on the state it stood in for — an element attached or visible, the editor's DOM reflecting the last keystroke, the document title reflecting the committed page title, or the updateBlockCollectionContents mutation's response. Sleeps that a following auto-waiting step already covered are deleted.

🔗 Related links

• Linear: FE-1578
• Prior work: FE-1577 / FE-1577: Poll for the property table cell value instead of sleeping #9526 (the entity-editing sleep), H-4863

🚫 Blocked by

• Nothing.

🔍 What does this change?

File	Was	Now
page-creation.spec.ts	500 ms before clicking the create-page button (both tests)	openHomeReadyToCreatePage (new shared helper): opens /, waits for the response to the sidebar's queryEntitySubgraph request for the web owner (useCreatePage only navigates to the new page once that shortname is known, and nothing in the sidebar renders from that query), then for pages-tree to be attached (addPage is a no-op while pages load)
	100 ms after the formatting shortcuts	expect(em).toHaveText("italics")
	100 ms after Enter	expect(blockHandles).toHaveCount(2)
	100 ms after typing /divider	suggester list item matching /divider/i visible
	100 ms after choosing the divider	deleted — the next steps already wait for the hr
	100 ms before typing the second paragraph	expect(blockHandles).toHaveCount(3)
	100 ms after typing text	expect(blockRegion).toContainText(...)
	100 ms after each Shift+Enter	expect(hardBreaks).toHaveCount(n) on br:not(.ProseMirror-trailingBreak)
	100 ms after the final Enter	deleted — toHaveCount(4) follows
	2 s "give collab time to sync" before reload	waitForResponse for the updateBlockCollectionContents response whose returned contents hold 4 blocks, armed before the final Enter
	3 × 500 ms before re-filling the page title	expect(page).toHaveTitle(...) with the committed name — the document title comes from the refetched page query, so it changes only once the saved title has propagated back to the page
page-readonly-mode.spec.ts	500 ms before clicking the create-page button	openHomeReadyToCreatePage, which also supplies the navigation this test was missing
inbox-page.spec.ts	2 × 12 s for the notification poll (default interval 10 s), with test.setTimeout(60_000) to fit them	page.reload() after creating each notification, then toBeVisible() with the default timeout — the notifications query is a network-only useQuery mounted by the page, so a reload fetches at once instead of waiting for the poll; test.setTimeout removed
profile-page.spec.ts	5 s for the bio editor to mount (×2)	placeholder text visible — the placeholder plugin renders it only when the editor is loaded, empty and focused
	2 s after "Save Bio" before reload (×2)	waitForResponse for the updateBlockCollectionContents request carrying the bio text, armed before typing
	2 s after reload (×2)	deleted — auto-retrying toBeVisible assertions follow

Left as they are: the poll interval in shared/get-kratos-verification-code.ts, the TOTP window wait in shared/totp-utils.ts, the debounce helper in extension/browser-plugin.spec.ts, and the sleep(500) in entity-editing.spec.ts, which #9526 removes.

playwright.config.ts is unchanged: no retries, no timeouts.

Pre-Merge Checklist 🚀

🚢 Has this modified a publishable library?

This PR:

• does not modify any publishable blocks or libraries, or modifications do not need publishing

📜 Does this require a change to the docs?

The changes in this PR:

• are internal and do not require a docs change

🕸️ Does this require a change to the Turbo Graph?

The changes in this PR:

• do not affect the execution graph

⚠️ Known issues

• Unverified by execution: the environment this was written in has no Docker, so the suite has not been run. CI is the first run.
• Three of the four specs (page-creation, page-readonly-mode, profile-page) are test.skip'd, so CI exercises only inbox-page.spec.ts.
• The block-count waits in page-creation.spec.ts (3 blocks after inserting the divider) take the counts the test already asserted as correct; they are not re-verified here.

🐾 Next steps

• If the create-page button gained a readiness state in the UI, the tests could wait on that instead of on the shortname query's response.

🛡 What tests cover this?

• The changed specs themselves; only inbox-page.spec.ts runs in CI.

❓ How to test this?

1. Run yarn test:integration in tests/hash-playwright against a local stack.
2. Confirm inbox-page.spec.ts passes in a few seconds rather than the 24 s it used to sleep.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

4 Skipped Deployments

Project	Deployment	Actions	Updated
hash	Ignored	Preview	Sep 3, 2026 6:39pm UTC
hashdotdesign-tokens	Ignored	Preview	Sep 3, 2026 6:39pm UTC
petrinaut	Skipped		Sep 3, 2026 6:39pm UTC
petrinaut-docs	Skipped		Sep 3, 2026 6:39pm UTC

[Copilot]

🟡 Changes recommended

Page creation needs a reliable shortname-readiness condition, and the inbox test timeout needs additional headroom.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

Replaces fixed Playwright sleeps with condition- and response-based waits to reduce test flakiness and runtime.

Changes:

• Adds DOM, title, visibility, and mutation-response waits.
• Removes obsolete fixed delays.
• Synchronizes notification and editor interactions with observable state.

File summaries

File	Review
tests/hash-playwright/tests/features/profile-page.spec.ts	Waits for editor readiness and bio-save responses; no issues found.
tests/hash-playwright/tests/features/page-readonly-mode.spec.ts	Waits for pages-tree, but this does not guarantee the separately queried owner shortname is ready before page creation.
tests/hash-playwright/tests/features/page-creation.spec.ts	Adds editor, title, and persistence waits; the page-tree wait can race with owner-shortname readiness and prevent navigation.
tests/hash-playwright/tests/features/inbox-page.spec.ts	Replaces polling sleeps with visibility assertions, but the 60-second test timeout lacks sufficient headroom for two 30-second waits plus setup.

Review details

• Files reviewed: 4/4 changed files
• Comments generated: 2
• Review effort level: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[cursor]

PR Summary

Low Risk
Test-only changes with no application or library behavior modified; CI impact is limited to integration test timing and stability.

Overview
Playwright specs under tests/hash-playwright no longer rely on fixed sleep(...) delays. Waits now track the UI or network state each pause was meant to cover, which should cut flaky failures on slow runners and wasted time when the app is fast.

A new shared helper openHomeReadyToCreatePage opens /, waits for the sidebar’s queryEntitySubgraph response for the web owner (so create-page navigation has the shortname), then for pages-tree to attach. Page-creation and read-only specs use it instead of a 500 ms delay before Create page.

inbox-page.spec.ts reloads after each API-created notification instead of waiting ~12 s for polling, and drops the 60 s test timeout. page-creation.spec.ts ties editor steps to DOM signals (block handles, italic text, divider suggester, hard breaks) and waits for a GraphQL updateBlockCollectionContents response with four blocks before reload; rename flows wait on toHaveTitle for the saved name. profile-page.spec.ts waits for the editor placeholder and for the bio save mutation in the request body.

Production code is unchanged; only test helpers and assertions.

^{Reviewed by Cursor Bugbot for commit 7223b79. Bugbot is set up for automated code reviews on this repo. Configure here.}

[claude]

Dequeued at 22:00 UTC by an unrelated Playwright failure in the queue run (run 33808993925): tests/extension/browser-plugin.spec.ts:256 "user can configure a one-off inference, and the settings are persisted" — the Actor chip never appeared (toHaveCount(1) got 0, then the 30 s test timeout). This PR does not touch that spec; the same head passed Playwright in its parallel queue group, on the PR branch, and main is green. This is the same flake that dequeued #9526 earlier today. Re-enabling auto-merge to re-enqueue.

---

Generated by Claude Code

[claude]

Dequeued again at 22:29 UTC by the same test, in run 33811421092: tests/extension/browser-plugin.spec.ts:256 "user can configure a one-off inference, and the settings are persisted" hit its 30 s test timeout while re-opening the popup after sign-out (openPopupTab at line 276, then page.evaluate: Target page, context or browser has been closed). Different failure point from the 22:00 run, same test, still a file this PR does not touch and a different Playwright project. This PR was alone in its queue group both times; its own CI on this head is green. Re-enabling auto-merge once more.

---

Generated by Claude Code