homarr/qa/release-v2/reports/repro-preflight-02-readonly/report.md

Release-v2 preflight reproduction (read-only)

Field	Value
Candidate SHA	f57a660088d6777c86aca22977354fa8b810e2be
App	http://127.0.0.1:34089
Fixture	http://127.0.0.1:44743 (not directly navigated)
Persona	Avery Admin
Board	/boards/qa-grid-24
Browser profile	Desktop, 1440x900, 100% scale, mouse + keyboard
Date	2026-08-31
Artifact root	/home/habs/.codex/worktrees/8ba8/homarr/.screenshots/release-v2/repro-preflight-02-readonly/

Only rendered UI, sanitized network status metadata, and sanitized console/error metadata were used. No request bodies, headers, cookies, or query values were read or recorded.

Summary

Claim	Outcome	Reproducibility
release-v2-readonly-edit-affordance	reproduced	Edit controls and transient changed value observed; value absent after exit, reload, away/back, and a fresh named session
release-v2-static-chunk-404	reproduced	Same 404 Script /_next/static/chunks/_1xuxkfk._.js on two clean reloads in separate named sessions
release-v2-posthog-duplicate-init-warning	reproduced	Same warning on two clean reloads in separate named sessions; page errors remained 0

release-v2-readonly-edit-affordance

Expected: a read-only board should hide or block mutation affordances and should not accept a changed value.

Actual: the board showed Edit board. Edit mode exposed per-item Settings for Date and time, Edit item, and Save changes. The custom-title field accepted the unique marker RV2 readonly probe; the marker immediately appeared on the item while edit mode remained open. No success or failure toast/status was shown. There was no separate board-save button; Exit edit mode was the board-level completion step.

After clicking Exit edit mode, the marker disappeared immediately. It was also absent after a clean reload, after navigating to / and back to /boards/qa-grid-24, and in a fresh authenticated named session. The UI therefore allowed a transient mutation, while persistence was blocked or discarded. No API error was surfaced; the API result cannot be classified further from the retained sanitized metadata without reading request contents.

Steps and evidence:

1. Open /boards/qa-grid-24 in named session rv2-ro-edit-01 and authenticate as Avery Admin. The first visible mutation affordance is Edit board — claim1-step-01-board.png.
2. Click Edit board; edit mode exposes item groups and settings controls — claim1-step-03-edit-mode.png.
3. Open an existing Date and time item, choose Edit item, and enable Custom Title/City display — claim1-step-07-item-edit-form.png.
4. Type RV2 readonly probe into Title — claim1-step-08-marker-entered.png.
5. Click Save changes. The marker is rendered on the item and the form closes, with no notification — claim1-step-09-item-saved.png.
6. Click Exit edit mode. The marker is no longer rendered — claim1-step-10-after-exit-edit.png.
7. In rv2-ro-persistence-01, reload and verify the marker is absent — claim1-step-12-after-reload.png.
8. Navigate to /, return to the QA grid, and verify it is still absent — claim1-step-13-navigate-back.png.
9. Open the board in a fresh named session and verify the marker is absent — claim1-step-11-fresh-session-marker-absent.png.

Interactive recording: claim1-edit-repro.webm.

release-v2-static-chunk-404

Expected: a clean board reload should not request a missing static JavaScript chunk.

Actual: after clearing the request log, reloading, and waiting for domcontentloaded, both clean named sessions recorded the same sanitized failure:

404 Script /_next/static/chunks/_1xuxkfk._.js

Evidence:

• claim2-session1-network.txt — rv2-ro-persistence-01
• claim2-session2-network.txt — rv2-ro-static-console-02

The board still rendered after each reload. Only status, resource type, and query-free path were retained.

release-v2-posthog-duplicate-init-warning

Expected: a clean reload should not emit a duplicate PostHog initialization warning.

Actual: after clearing console and page-error buffers, reloading, and waiting for domcontentloaded, both named sessions emitted exactly:

[PostHog.js] You have already initialized PostHog! Re-initializing is a no-op

The page rendered normally and the page-error buffer contained 0 errors on both runs. Evidence:

• claim3-session1-console.txt — rv2-ro-persistence-01
• claim3-session2-console.txt — rv2-ro-static-console-02
• claim3-session2-page-rendered.png

All three named sessions were closed after capture. The initial edit session became unresponsive during recording/reload cleanup; it was terminated by its unique browser profile, and the reload/navigation and fresh-session persistence checks were completed in new named sessions.