Gutenberg: v23.2.x plugin upgrade

[gutenberg-plugin-upgrade-bot]

Previous upgrade: #110322
Release notes: v23.2.1

Install and Activate

• Install and activate v23.2.0-rc.1 on edge (217742-ghe-Automattic/wpcom)
• Install and activate v23.2.0 on edge (######-ghe-Automattic/wpcom)
• v23.2.0 WPCOM Core unit tests (217693-ghe-Automattic/wpcom)
• Activate v23.2.0 on production (######-ghe-Automattic/wpcom)
• Install and activate v23.2.1 on edge (219199-ghe-Automattic/wpcom)
• v23.2.1 WPCOM Core unit tests (219200-ghe-Automattic/wpcom)
• Activate v23.2.1 on production (219200-ghe-Automattic/wpcom)

Wrap-up

• Clean up unused releases
• Add log entry to the Gutenberg Upgrade Log (pcoGjb-g-p2)

Blockers 🤷‍♀️

Other Issues 🐛

Issues transferred from previous release(s)

---

cc @Automattic/gutenberg-upgrades

[gutenberg-plugin-upgrade-bot]

Automated update:

• v23.2.0-rc.1 activation on edge marked as complete.
• Release notes updated to v23.2.0-rc.1
• Updated the PR used for v23.2.0-rc.1 core tests

[xavier-lc]

Flake fix for editor__post-advanced-flow.ts → Switch to draft (TC build 17776184, GB v23.2.0-rc.1): #110795

The test waited on the "Post reverted to draft" snackbar dismiss button, which auto-dismisses after ~5s; the PR replaces it with a wait on the toolbar primary button label reverting Update → Publish, which only commits once the draft transition has persisted.

[xavier-lc]

Fix PR for the mobile flake on specs/blocks/blocks__coblocks__extensions__replace-image.ts:57 ("CoBlocks Replace Image: Insert Image block and upload image"): #110797

Root cause: on mobile the inserter sometimes auto-closes after insertion; the "Close Block Inserter" button could pass actionability checks at the start of closeBlockInserter()'s click and then detach mid-animation, hanging the click until the action timeout. The fix early-returns when the sidebar is already gone and otherwise retries the click up to 3 times with a bounded per-attempt timeout — sidebar detachment is the success oracle. Verified with three consecutive green local runs on GUTENBERG_EDGE=true VIEWPORT_NAME=mobile.

[xavier-lc]

Guidelines: Fix fatal when rest_api_init fires before init

Fixed on GB repo.

[gutenberg-plugin-upgrade-bot]

Automated update:
Queued e2e test builds for v23.2.0 on edge:

• Atomic desktop e2e tests: https://teamcity.a8c.com/build/17840581
• Atomic mobile e2e tests: https://teamcity.a8c.com/build/17840580

[gutenberg-plugin-upgrade-bot]

Automated update:
Added tasks for v23.2.1

[gutenberg-plugin-upgrade-bot]

Automated update:
Queued e2e test builds for v23.2.1 on edge:

• Atomic desktop e2e tests: https://teamcity.a8c.com/build/17850164
• Atomic mobile e2e tests: https://teamcity.a8c.com/build/17850163

[gutenberg-plugin-upgrade-bot]

Automated update:

• v23.2.1 activation on edge marked as complete.
• Release notes updated to v23.2.1
• Updated the PR used for v23.2.1 core tests

[gutenberg-plugin-upgrade-bot]

Automated update:
Queued e2e test builds for v23.2.1 on edge:

• Simple desktop e2e tests: https://teamcity.a8c.com/build/17864296
• Simple mobile e2e tests: https://teamcity.a8c.com/build/17864297

[xavier-lc]

Heads-up on stale mutes during v23.2.1 e2e (not blocking this rotation, FYI for QA)

While walking the e2e suites for v23.2.1, I noticed 6 muted test failures across the 4 edge e2e builds where muted: true (so they didn't fail the build) but currentlyMuted: false at the project level. That means: the next time these tests run, the global mute won't shield them and the build will fail. None of these look like v23.2.1 regressions — they're classic UI-selector flakes / pre-existing infrastructure issues — but flagging so QA can re-mute or fix before they bite the next rotation.

Simple edge desktop (TC #17864296) — today

• specs/editor/editor__page-basic-flow.ts — "Editor: Basic Page Flow: Select page template" — locator.waitFor: Timeout 15000ms exceeded waiting for 'Block pattern "About 12" inserted.' snackbar (at editor-page.ts:529 → addPatternFromSidebar → editor__page-basic-flow.ts:81).
• specs/editor/editor__post-advanced-flow.ts — "Revert post to draft: Re-open the published post from the Posts page" — locator.waitFor: Timeout 20000ms exceeded for tr.type-post a.row-title:has-text("Post Life Cycle: …") (at posts-page.ts:109 → ensurePostShown → clickPost → editor__post-advanced-flow.ts:152).

Simple edge mobile (TC #17864297) — today

• specs/editor/editor__post-advanced-flow.ts — same "Re-open" timeout as above.

Atomic edge desktop (TC #17850164) — yesterday

• specs/editor/editor__post-advanced-flow.ts — "Switch to draft" — TypeError: Cannot read properties of undefined (reading 'unicode') at playwright-core/lib/utils/isomorphic/stringUtils.js:82 in escapeRegexForSelector (Playwright internal — likely a Playwright version mismatch, not editor code).
• specs/blocks/blocks__coblocks__extensions__replace-image.ts — "Insert Image block and upload image" — timeout 15s waiting for img:not([src^="blob:"]) (at image-block.ts:59).
• specs/blocks/blocks__jetpack-forms.ts — "Text input field: Configure the block" — locator.click: Timeout 15000ms for 'Select parent block: Form' button (at editor-block-toolbar-component.ts:146).

The 2 likes__comment.ts "Setup the test" failures (RestAPI Failed to like N on site 202432195) on the atomic suites are still globally muted (currentlyMuted: true) and so don't have this issue — flagging only to acknowledge they were also seen.