LearnNewsExamplesServices
Frontmatter
id18037
titleExplain how to test and debug DockLayouts end to end
stateClosed
labels
documentationenhancementdeveloper-experienceno auto closeaitestingcore
assigneesneo-opus-ada
createdAtSep 1, 2026, 7:52 PM
updatedAtSep 13, 2026, 12:07 PM
githubUrlhttps://github.com/neomjs/neo/issues/18037
authorneo-gpt-emmy
commentsCount0
parentIssue17540
subIssues[]
subIssuesCompleted0
subIssuesTotal0
contentTrust
projected
quarantined0
signals[]
blockedBy[x] 17836 Rebuild DockLayouts for the v13.2 greenfield architecture
blocking[]
closedAtSep 13, 2026, 12:07 PM

Explain how to test and debug DockLayouts end to end

Closed v13.2.0/archive-v13-2-0-chunk-17 documentationenhancementdeveloper-experienceno auto closeaitestingcore
neo-gpt-emmy
neo-gpt-emmy commented on Sep 1, 2026, 7:52 PM

Context

DockLayouts has an unusually deep evidence surface: pure model units, component fixtures, headed whitebox E2E, Neural Link worker/main inspection, rendered theme checks, mutation controls, and external-Brain runtime gates. The generic testing guides cannot encode the subsystem-specific evidence boundaries.

The Problem

Maintainers repeatedly get green-looking wrong answers by sampling the wrong plane, using stale theme output, skipping the external runtime gate, observing live DOM instead of VDOM ownership, or treating a test tally as failure disposition. There is no guide explaining how to choose an instrument that can contain the DockLayouts answer.

The Architectural Reality

The current suite spans test/playwright/{unit,component,e2e}/dashboard, Workstation whitebox journeys, theme fixtures, and Neural Link. Model truth, App-worker component truth, VDOM delivery, painted DOM, and physical windows are distinct observation surfaces; custom Playwright configs and headed execution are mandatory for different questions.

The Fix

Create learn/guides/testing/DockLayoutsTesting.md: an evidence-first guide to reproducing, observing, mutating, and debugging DockLayouts across its model, worker, DOM, and window planes.

Contract Ledger

Target Surface Source of Authority Proposed Behavior Fallback Docs Evidence
Testing guide Current DockLayouts unit/component/E2E suites + Neural Link Teach instrument choice by authority plane and failure class Re-census suite at pickup New guide Red-first and mutation receipts
Runtime gates Current Playwright configs + external Brain root behavior Explain when a listed test is actually discoverable/executable Fail closed on missing runtime Same guide List/run controls
Registration Existing testing-guide siblings One discoverable Testing guide No generated-output edits tree + SEO inputs lint + rendered portal

Decision Record impact

aligned-with ADR 0029

Decision Record

Not needed; this guide explains evidence collection.

Acceptance Criteria

  • The guide clears the full guide-authoring bar: source/tool-grounded narrative, reader-owned benefits, lived maintainer voice, and at least one render-verified flowchart TD.
  • The guide separates explanation from reference: no inlined API/schema/test catalog, and no ticket or PR ids in guide prose.
  • The new source guide is registered in learn/tree.json and the SEO PRIORITIES input; generated sitemap/LLM outputs remain untouched.
  • Current-source consistency, guide lint, tree lint, portal rendering, and an independent cross-family whole-guide review are recorded in the PR.
  • The guide maps model, App worker, VDOM, DOM/paint, main-thread geometry, and physical-window questions to instruments that can observe them.
  • It uses current custom unit/component/E2E configurations and demonstrates at least one red-first reproduction plus a mutation control.
  • It explains external-Brain runtime discovery, headed requirements, theme freshness, deterministic versus batch-order failures, and privacy-safe diagnostics.
  • It links generic testing and Neural Link references instead of copying commands or tool catalogs.

Out of Scope

A full test inventory, CI redesign, or fixing current DockLayouts defects.

Avoided Traps

Do not run default Playwright, claim zero tests as green, use fixed sleeps as semantic evidence, or collapse worker truth and painted DOM into one plane.

Activation Gate

Do not author the guide while #17836 remains an active refactoring authority. At pickup, re-run Memory Core mining, read the current ADR 0029 and source/test surfaces named below, exercise the relevant live subsystem, and revalidate the proposed guide boundary. If the final API makes this split wrong, reshape this ticket before writing—do not preserve a stale outline.

Related

Related: parent guide Epic #17540; blocked by active DockLayouts refactor Epic #17836; authority ADR 0029.

Origin Session ID: 309a78d7-2d95-4f23-bdfc-69d2ae393a5d

Retrieval Hint: query_raw_memories("DockLayouts guide series post-refactor final API")

Creation freshness: live created-descending sweep of the latest 20 open Engine issues at 2026-09-01T17:49:38Z found no equivalent guide ticket; the all-state latest-30 A2A sweep found no overlapping guide claim. The Knowledge Base returned stale pre-cut names, so live source, ADR 0029, and the current guide tree remain authority.