LMarbles Original Puzzle 2: native clear, 1650 tally, Puzzle 3
Fresh registered lmarbles_108, Original, Tier 1 Puzzle 2 cleared through
fifteen ordinary, individually reviewed slides. The native completion particle
animation displayed Moves 13. The subsequent tally displayed Total Score
1650, with both bonus counters depleted to zero. Ordinary Enter advanced to
Tier 1 Puzzle 3, Moves 34, Score 1650. These are current runtime observations,
not a promotion of the historical Moves 16 checkpoint or static route.
Evidence: scratch/runs/20261114-lmarbles-puzzle2-clear/result.json. Personally
reviewed retained full native frames:
board.png: Original Tier 1 Puzzle 2, Moves 28, Score 0.before-final.png: Moves 14, Score 0, selected red at (3,3), red at (5,3), maroons at (4,4)/(6,4), yellow at (5,4).clear-animation.png: native marble particle explosion, Moves 13.tally.png: Total Score 1650 over the completed Puzzle 2 board.next.png: Original Tier 1 Puzzle 3, Moves 34, Score 1650.
No literal “CLEAR” caption was observed. The native completion animation, authentic tally, score award and next-puzzle transition establish the clear.
The fresh menu retained Normal/Original/Michael defaults. Hovering the second gold selector visibly labelled Tier 1 Puzzle 2; clicking produced the native “enter level 2 of chapter 1?” confirmation. Balanced Y entered the fresh board. No Tutorial completion was used. The board and target were inspected before gameplay: the target was red above yellow, over red–maroon–yellow–maroon.
Public-frame tile centers were (98+32*x,154+32*y). Each selection ring and
every settled consequence was personally inspected before the next input.
Where the same marble remained selected, its retained ring was reviewed
without an unnecessary second click.
| Slide | Color | From → To | Arrow | Moves left |
|---|---|---|---|---|
| 1 | red | (8,7) → (8,1) | Up | 27 |
| 2 | maroon | (6,1) → (4,1) | Left | 26 |
| 3 | red | (8,1) → (5,1) | Left | 25 |
| 4 | yellow | (1,7) → (2,7) | Right | 24 |
| 5 | yellow | (2,7) → (2,4) | Up | 23 |
| 6 | maroon | (4,4) → (8,4) | Right | 22 |
| 7 | yellow | (2,4) → (7,4) | Right | 21 |
| 8 | maroon | (4,1) → (4,4) | Down | 20 |
| 9 | red | (3,1) → (4,1) | Right | 19 |
| 10 | red | (4,1) → (4,3) | Down | 18 |
| 11 | red | (4,3) → (3,3) | Left | 17 |
| 12 | yellow | (7,4) → (5,4) | Left | 16 |
| 13 | maroon | (8,4) → (6,4) | Left | 15 |
| 14 | red | (5,1) → (5,3) | Down | 14 |
| 15 | red | (3,3) → completion | Down | 13 |
The final expected stop was (3,4); the captured consequence was the native completion animation rather than a persistent five-marble settled frame. The historic three-slide hints were used only after this run's preceding states and rings were freshly reviewed.
Current source was built successfully under both /tmp/wine-heavy-runtime.lock
and /tmp/wine-build.lock. Launch source commit was
3f6b5694ab1e7cf048aa29881221a2250aa5019d, plus the recorded relevant binary
dirty diff and untracked-runtime patch. Loaded module SHA-256 was
c7f5bada4da4b4ed95b235add0aa1c9843176805d29d5b1ee79359752a19d348.
source-pins.json, runtime-dirty.patch, runtime-new.patch and
owned-helpers.patch preserve source identity using Git, without source
archives or module dumps. The module remained unchanged through shutdown.
Only lib/d3d9-render-worker.js changed among the launch-pinned runtime files
during gameplay, by its separate owner; this cooperative DirectDraw game did
not launch that D3D9 worker. No foreign WIP was adopted or committed.
Original EXE remained at
test/binaries/candidates/lmarbles-108/game/LMARBLES/lmarbles.exe, SHA-256
574df6673f136d399ab16afc145ca1ebf2589772884f2f47ce18c016f0fa40c5.
All existing 86 original media files plus CRT matched their supplied pins both
before and after execution. The registered manifest/CWD were used directly;
media was neither copied nor restored. Runtime used real ticks, cooperative
threads, a 1024×712 canvas and the ordinary control channel.
The new scratch controller required a fresh hashed native PNG and explicit observation/expected consequence for each action. Each DOWN had an immutable expiry before an independently prearmed release. All 32 presses had release children actually joined at exit 0 with both EOFs and subsequent native batch progress. All 34 actions and fifteen slides passed the postflight checks. An initial motion-only hover was released too early to reach the guest; a controller-only delivery delay produced the reviewed hover. Handler consumption was never treated as proof of guest input delivery or visual success.
Focused tests test-cli-guarded-input.js and
test-control-server-step-progress.js passed; both scratch Python helpers
passed syntax parsing. No production source repair was needed. Only this
document is committed. Intermediate reviewed PNGs were overwritten during
operation and removed at the end; only the five result-named key PNGs remain.
Compact action records preserve their hashes, observations and receipts.
Independent watchdog PID3196469 acknowledged its finite 1500-second deadline before native PID3197737 existed. Runtime also had a 1300-second limit; neither deadline was extended. After the next-puzzle capture, normal control quit returned native exit 0; stdout/stderr were drained to EOF and the process was absent. Watchdog exit 0, board watcher TERM/wait, and outer session59473 exit 0 were all observed. The heavy lock was released before evidence finalization. Box ID none; state NOT_CREATED; expiry null; stop receipt N/A.
The assigned clear/tally/continuation goal is fulfilled. No guest/configuration/ clock/score writes, state import, forced readiness, fake acknowledgments, browser, public deployment, other workers or boxes were used. Further Puzzle 3 gameplay would be a separate task.