--- phase: 06.1-gap-closure-d-39-d-46-uat-2026-05-11 status: pending-rerun last_updated: 2026-05-12 operator: decidel re_runs_after_findings: - D-39 - D-40 - D-41 - D-42 - D-43 - D-44 - D-45 - D-46 --- # Phase 06.1 — Human UAT (re-run after gap-closure) > Five manual verifications targeting the 2026-05-11 finding set (D-39..D-46). Test 2 (CLI-08) is THE milestone artifact for the rebuild — capture `06.1-CLI-08-milestone.mp4` if all five pass. The 2026-05-11 video at `.planning/phases/06-client-rebuild-mvp-gate-cli-08-hard-milestone/uat-test-1-2.mp4` is the ground-truth broken-state reference. This script follows the format of `.planning/phases/06-client-rebuild-mvp-gate-cli-08-hard-milestone/06-HUMAN-UAT.md`. Run AFTER all Phase 06.1 plans (06.1-01..06.1-07) merge and the staging CI auto-redeploys. Verify deploy: visit `https://staging.rebno.decidel.com/?invite=$STAGING_INVITE_TOKEN` and confirm the build hash in the page footer matches the latest `main` commit. --- ## Test 1 — Alpha-keyed sprites visible (D-39 closure) **Why manual:** Background+sprite layering on real GPU with real CSP / blob: img-src wiring is not headless-reproducible at staging-equivalent fidelity. Closes D-39 (solid white box behind sprites). **Why it failed 2026-05-11:** BMP→PNG bootstrap did not key the source-white pixel to alpha; sprites rendered with a solid white halo per the 2026-05-11 video. **Instructions:** 1. Open `https://staging.rebno.decidel.com/?invite=$STAGING_INVITE_TOKEN` in Chrome. 2. Log in as `uat_a`. Wait for GameScene (`data-game-ready="true"`). 3. Take a 4× zoom-in screenshot of the local NaviStandD sprite over the mvp-room floor tiles. 4. Inspect the sprite outline: no opaque magenta, no white halo, no solid rectangle behind the figure. Background tiles must be visible AROUND the sprite, including in the gaps between the legs / between the head and torso. 5. Save the screenshot as `.planning/phases/06.1-gap-closure-d-39-d-46-uat-2026-05-11/06.1-test-1-alpha.png`. **Pass criteria:** - No solid color box (white or magenta) around the sprite. - Floor tiles visible in transparent regions of the sprite bbox. - Aseprite-roundtrip (read-only): open `assets/source/sprites/0000-NaviStandD.png` in Aseprite — file opens, alpha channel present, no white-edge artifacts. **Fail signals:** - Solid white box around sprite → D-39 fix did not land or atlas was not regenerated by CI. - Magenta sentinel visible → bootstrap keyed to wrong sentinel RGB. --- ## Test 2 — Two-player CLI-08 MILESTONE (REQ-CLI-08) **Why manual:** The "we shipped CLI-08" milestone artifact requires human capture. Playwright (the four `cli-08-*.e2e.test.ts` files from Plan 06.1-03) covers regression; this is the operator deliverable. **Why it failed 2026-05-11:** Networking + protocol works; visuals (D-39..D-46) blocked closure. **Instructions:** 1. Operator opens 2 Chrome browser windows (or two physical machines) against `https://staging.rebno.decidel.com/?invite=$STAGING_INVITE_TOKEN`. 2. Sign in as `uat_a` in window 1 and `uat_b` in window 2. 3. Wait for both clients to enter the `mvp-room` GameScene (`data-game-ready="true"` AND `data-room-id="mvp-room"`). 4. Confirm visually in both windows: floor tiles (`0023-Tile1` grid) visible, TSide1 bottom edges visible along the outer floor row, animated `0064-BKA1` background scrolling slowly down-and-right behind/around the room. 5. Start a screen recording covering BOTH browser windows (OBS / Loom / built-in screen recorder). 6. In window 1 (uat_a): walk toward uat_b using WASD (~5 seconds of movement). Confirm: local NaviRun animation cycles smoothly (≥ 3 distinct frames during a 1-second hold), remote sprite (uat_b) animates at the same visual cadence. 7. In window 2 (uat_b): walk toward uat_a. 8. Confirm visually: each window shows BOTH player sprites + nameplates moving smoothly; local nameplate is cyan c_aqua, remote nameplate is white c_white; both nameplates float above sprite head with a 1 px black shadow at offset (1, 1). 9. In window 1 (uat_a): press Enter, type `hello uat_b`, press Enter. 10. Confirm: window 2 shows `uat_a: hello uat_b` in the chat HUD within 2 seconds. 11. In window 2 (uat_b): press Enter, type `hi uat_a`, press Enter. 12. Confirm: window 1 shows `uat_b: hi uat_a`. 13. Stop the recording. Save as `.planning/phases/06.1-gap-closure-d-39-d-46-uat-2026-05-11/06.1-CLI-08-milestone.mp4`. **Pass criteria:** - Both clients see each other's sprite + nameplate. - Bidirectional chat works in real time. - Floor + TSide1 + animated background all render. - Movement feels smooth (no choppiness; verified subjective + by Test 3/4 below). - Video file exists and is committed (or linked). - Operator types "approved — CLI-08 shipped" in the resume signal. **Fail signals:** - Either client sees only blue rectangles → 06.1-02 atlas regen did not deploy. - No floor tiles → D-40 not closed (re-check Plan 06.1-04 verify-bail flip landed). - Chat HUD missing after movement → re-entry regression (Phase 6 D-25 reopened). --- ## Test 3 — Anim rate consistency (D-41/D-42 closure) **Why manual:** Local-vs-remote animation cadence symmetry is a subjective visual call beyond what a frame-count assertion captures. **Why it failed 2026-05-11:** D-41 (local stuck) + D-42 (remote too fast) both rooted in the sim-tick callsite contract being lost between Phase 6 plans 06-16 and 06-17. **Instructions:** 1. Continue from Test 2 (both clients still logged in to mvp-room). 2. In window 1 (uat_a): hold KeyD for exactly 2 seconds. Visually count the run-cycle frame transitions on the LOCAL sprite (should see ~15 frame advances across the 6-frame cycle = ~2.5 full cycles, i.e. the visible run animation completes 2-3 loops). 3. In window 2 (uat_b): observe uat_a's REMOTE sprite during the same 2-second window. The remote sprite must animate at the same visual cadence — NOT faster, NOT slower. 4. Repeat with uat_b holding KeyA and uat_a observing. **Pass criteria:** - Local + remote run animations cycle at visually identical rate (within human-eye tolerance). - Neither is "stuck" (frame visibly advances on every keypress). - Neither animates faster than ~10 fps visual (5 fps in the canonical 0.5 frames/tick × 30 Hz model = 15 fps real-time, fast-but-not-jarring). **Fail signals:** - Local sprite stuck → D-41 not closed (PlayerRenderer.onSimulationTickLocal not firing). - Remote sprite at 60Hz blur → D-42 not closed (caller still on Phaser render frame). - Remote slower than local → asymmetric sim-tick wiring; check PlayerRenderer.onSimulationTickRemote callsite. --- ## Test 4 — Camera follow smoothness + nameplate legibility + Fixedsys font load (D-45/D-46 closure) **Why manual:** Subjective camera-feel + font legibility + nameplate position calls. **Why it failed 2026-05-11:** D-45 (nametag too large/wrong color) + D-46 (camera does not follow) both visible in `uat-test-1-2.mp4`. **Instructions:** 1. Continue from Test 2/3 (both clients still in mvp-room). 2. In window 1 (uat_a): walk continuously in a circle for ~10 seconds (W → D → S → A → W). 3. Observe: camera tracks the local sprite — the sprite stays in roughly the center of the viewport (within the D6.1-14 32×32 deadzone). No "leaving the screen" sensation. 4. Observe: as the camera scrolls, the animated `0064-BKA1` background continues to scroll under/past the room edges — no "void" at the room boundary; world feels infinite. 5. Open Chrome DevTools (F12) → Network tab → reload page → confirm `Fixedsys` (or whatever the chosen Fixedsys-Excelsior WOFF2 is named under `apps/client/public/assets/fonts/`) loads with HTTP 200, MIME `font/woff2`. 6. Visually inspect both nameplates: cyan (local) and white (remote), small pixel font (~10 px), positioned 16 px above sprite head, 1 px black drop-shadow at offset (1, 1). **Pass criteria:** - Camera holds local sprite within the central deadzone area for the full 10-second loop. - Background tiles past the room edge — no black void. - Fixedsys WOFF2 loads cleanly (no fallback to system monospace). - Local nameplate cyan c_aqua, remote nameplate white c_white. - Nameplate has a visible drop shadow on light tiles. **Fail signals:** - Sprite walks off-screen → D-46 not closed (camera startFollow not re-bound to local sprite). - Black void beyond room edge → D6.1-16 unbounded camera + bkdraw not wired (Plan 06.1-05 / 06.1-06 not landed). - Generic monospace font in nameplate → WOFF2 not loaded (CSP or path issue). - Same color on both nameplates → D6.1-11 color rule not implemented. --- ## Test 5 — Walkable-edge block + background scroll/visible past room edge (D6.1-28/29 closure) **Why manual:** "Player blocked at edge" + "background visible past edge" are visual / haptic checks. **Why it failed 2026-05-11:** Not exercised — D-40 (no tiles) made the test impossible to run. Re-run now that tiles render. **Instructions:** 1. Continue from Test 4 (both clients in mvp-room). 2. In window 1 (uat_a): walk to the bottom edge of the floor region (the southernmost row of `0023-Tile1` sprites). Continue holding KeyS. 3. Observe: local sprite stops at the floor edge; cannot walk off into non-floor space. The NaviMask feet bbox (18×7 px) sits flush against the floor's south edge. 4. Observe: while blocked at the edge, the animated `0064-BKA1` background continues to scroll behind/around — the world "outside" the floor is visible and lively, not a black void. 5. Walk in each of the other three cardinal directions to the corresponding floor edges; confirm the block applies at all four edges. 6. Walk diagonally into a corner (e.g. SE) and confirm wall-slide: the sprite slides along the free axis when one axis is blocked (per D6.1-28 per-axis sub-pixel collision). **Pass criteria:** - Sprite blocks at all four floor edges with NaviMask feet bbox at the boundary. - Background scrolls past the room edge — infinite-feel. - Diagonal-into-corner produces a wall-slide along the unblocked axis. **Fail signals:** - Sprite walks past floor edge → walkable-region mask not derived OR per-axis collision not wired (Plan 06.1-05). - Black void past edge → D6.1-16 unbounded + bkdraw not landed (Plan 06.1-06). - Diagonal-into-corner produces a hard stop (no slide) → D6.1-28 per-axis sub-pixel loop not landed. --- ## Sign-Off (after re-run) | Test | Requirement | Status | Notes | |------|-------------|--------|-------| | 1 — Alpha-keyed sprites | REQ-AST-01, REQ-CLI-06 | ⬜ pending | run after 06.1-01 deploys | | 2 — CLI-08 milestone | REQ-CLI-08 | ⬜ pending | run after 06.1-01..06.1-06 deploy | | 3 — Anim rate consistency | REQ-CLI-04 | ⬜ pending | requires 06.1-04 PlayerRenderer fix | | 4 — Camera + nameplate + font | REQ-CLI-05, REQ-CLI-06 | ⬜ pending | requires 06.1-04 + 06.1-05 | | 5 — Walkable-edge + background | REQ-CLI-07 | ⬜ pending | requires 06.1-05 + 06.1-06 | **Operator approval pending.** Update this section after the re-run. **Resume signal:** Operator types one of: - `approved — CLI-08 shipped` → close Phase 06.1, archive `06.1-CLI-08-milestone.mp4`, advance to Phase 7. - `not approved — ` → spawn `/gsd-plan-phase 6 --gaps` for a third cycle. --- ## Re-Run Procedure Summary 1. Wait for CI to redeploy after Phase 06.1 plans merge to `main` (deploy-staging.yml auto-fires). 2. Verify the deployed build hash matches `main` HEAD. 3. Run Tests 1–5 in order (Test 2 is the milestone capture; Tests 3/4/5 piggyback on the same two-client session). 4. Run Plan 06.1-03's four Playwright e2e files (`pnpm --filter @rebno/client test:e2e -- cli-08-camera cli-08-tiles cli-08-nameplate cli-08-anim`) — all must exit 0 (GREEN gate alongside human UAT). 5. Commit the `06.1-CLI-08-milestone.mp4` + `06.1-test-1-alpha.png` artifacts to this phase directory. 6. Update the Sign-Off table above and append the operator's verdict line. --- *Phase 06.1 UAT re-run script. Authored 2026-05-12 by Plan 06.1-03 executor. Operator must run after all Phase 06.1 plans land.*