TIM earned route and early Quit, 2026-11-15
Result: native Quit proved at 220.332 seconds; Puzzle 2 not reached. The one authorized fresh original reached the ordinary Puzzle 1 construction board. Every construction attempt was positively refused before delivery; the unchanged five-second review / one-second dispatch margin could not admit this worker's observed image-review latency. There is no new construction, Run, clear, earned password or authored Puzzle 2 boundary claim.
Evidence: scratch/runs/20261115-tim-earned-route-early-quit-puzzle2/result.json.
This task authorizes one fresh unchanged installed original, with a 900-second
native deadline and independent 920-second watchdog. Construction stops by
elapsed 600 seconds; panel/Quit begins immediately, targeting actual native
exit by 720 seconds. Historical closed UNKNOWN runs remain closed and UNKNOWN.
The retained public completion screen in
20261115-tim-schema-checked-belt-progress/result-puzzle1-completed-bonus626.png
was personally viewed. It says PUZZLE 1 COMPLETED, bonus 626, and displays a
New Password. An initial transcription as SIERRA-ZZZZSIM was overconfident:
closer visual inspection leaves one glyph ambiguous. It will not be entered
without resolving the exact public lettering. The adjacent retained Replay/Advance dialog and Puzzle 2 panel
were also personally viewed. They establish an earned ordinary route, not
current progress. No password table, private game state, cheat list or config
was read. The unchanged route admits Enter/Escape/Space and mouse gestures;
it has no password typing operation. The coordinator's same-task clarification
on the messageboard at 09:17:10, acknowledged at 09:18:36, authorizes narrow
finite single-key support after the heavy release, prioritizing this genuine
earned password over rebuilding Puzzle 1 solely because of that allowlist.
The prepared, unapplied extension admitted individual uppercase letters and hyphen, with
the same exact schema, personal scene review, finite hold, independent UP and
native ambiguity rules. It supplies no string queue or automatic typing.
The actual chooser result must establish whether an exactly identified
displayed password unlocks Puzzle 2. An ambiguous or rejected password is
never guessed around. Ordinary manual Puzzle 1/Continue/Advance remains the
authorized fallback. The narrow-key candidate has only been tested in memory;
no shared controller change or native key test ran during the freeze. The
optional transcription remained unresolved, so the proposal was not applied;
the new test proposal was removed after preserving its patch. This original
uses the unchanged published mouse/special-key route and manually solves P1.
The original scanned manual was personally viewed at PDF pages 12, 14, 15 and 20 (printed 10, 12, 13 and 16). Its authored basketball mechanism is a cascade: a falling ball starts the first mouse motor, whose belt drives the first conveyor; each moving bowling ball starts the next motor/conveyor; the last conveyor rolls the basketball onto a ramp toward the hoop. Placement requires a separate drop click. A belt requires two attachment clicks, with the red line stretched between them. A handle click flips a placed object. These rules and each new complete scene determine actions; historical input coordinates and prerecorded action sequences are not replayed.
The retained public Puzzle 2 panel says to put both bowling balls into metal baskets, place and flip the mouse cage, then hook its pulley belt to the conveyor. The retained complete changed-cage scene shows the right pulley moved to the left. Current placement, direction, endpoints and Run consequences must each be established anew in the fresh original.
Quit is prepared before play: the manual explicitly identifies right mouse button as return to the native control panel and its red octagon as Quit. Use a new reviewed complete scene for each step, valid button 2 for return and button 1 for the visible Quit control, with an explicitly selected finite 80–1400 ms hold and independently enforced UP within 1750 ms. Clicking Quit or receiving a transport acknowledgment does not prove guest exit: the actual native machine exit and its consequence must be read separately from host closure. Written UNKNOWN permanently ends input. Native-attested positive UNSENT alone permits another complete capture, review, nonce and independent intent. Gameplay reviews remain five seconds; menu/Quit reviews thirty.
The coordinator's owner-priority update at 09:26:01 changed the remaining heavy order to Q3compact761864, fresh Sims818667, then this task; this worker acknowledged it at 09:26:54. Sims677879 already released. No shared source/controller/helper is changed while predecessors hold their freeze. Their explicit releases, direct native/watchdog/supervisor-child actual waits with both EOFs, and current PID/group absence must be personally verified before current qualification or the original. No box or worker has been created.
Sims818667 was replaced by Sims885263, acknowledged at 09:45:12. Q3compact released at 09:45:24 and Sims885263 at 10:09:15. Before qualification this worker personally read the latter's native/watchdog/supervisor-child/launcher actual waits (all zero, both EOFs), verified its 609 sealed waits and 823 absent PID/groups, and reverified Q3's final 108 waits and 345 absent PID/groups. Historical ancillary UNKNOWN records remain unchanged. Shared runtime is released to this task, followed by Q3layout916770.
Current qualification passed: full locked build; 26 policy, 12 exact-schema and 10 dispatch-margin tests; native mouse callback/context and held-key checks; exact CLI/native schema; both native dispatch-margin cases; and all 13 native controller cases (hover, click, right, drag, cancel, watchdog, parent death, unsent, late/drop ACK, held input, written stale, Quit reserve). All 96 named actual waits have both EOFs; 99 recorded PIDs/groups are absent. Two run-wrapper cleanup mistakes referenced the discarded key proposal after the build/tests passed; these stopped the wrapper before the remaining tests, were corrected, and remaining tests passed without rebuilding or changing the controller. No original was used for qualification. Its new native deadline watchdog must be visibly armed before launch. Source/build/module remain frozen throughout that original, and any written UNKNOWN ends input forever.
The fresh original started on published recipe e3c695f90, using the unchanged
controller from 712384ad6. Its independent watchdog was visibly armed before
native launch, with 900 native / 920 watchdog seconds. The current ordinary
challenge asked for manual page 1: personally viewed fishbowl, cat and bulb
icons were selected individually, leading to the Puzzle 1 panel and Play.
The uncertain historical password was never entered. Each game attempt had
a fresh complete scene and a new independent intent. Eight positive UNSENT
refusals had review ages 6.839, 8.197, 5.176, 4.390, 4.247, 4.485, 4.575 and
4.339 seconds; each has its own native zero-send attestation. Even the last
five were outside the four-second admission window after preserving the
one-second dispatch margin. No construction DOWN reached the game. This is
a measured interaction boundary, not evidence that the authored puzzle or
ball/motor/ramp mechanism is impossible. The prepared narrow-key proposal
remains unapplied, and no engine change was made.
Construction ended early. A newly reviewed right-button return reached the
native control panel at elapsed 211.1 seconds. Its visible red octagonal Quit
opened the native QUIT GAME confirmation at 216.0 seconds. A newly reviewed
YES click used explicit button 1 and a 100 ms hold. Its independent release
completed; native-final.json then recorded nativeExited: true,
nativeExitCode: 0, elapsed 220332 ms, held false and no game-state writes.
There was no host cancellation or watchdog KILL in this original. This is
the actual Quit consequence, not an ACK, fixture, host shutdown or default
exit-code inference. The guest's exit changed native ownership before the
controller could record its post-release consequence, so the controller
wrote permanent UNKNOWN (native owner changed). Native host process exit
1 and controller UNKNOWN remain unchanged; guest exit 0 is a separate fact.
No input, capture request, replay, rebind or second original followed UNKNOWN.
Direct native / deadline-watchdog / supervisor-child / launcher actual waits
were 1 / 0 / 1 / 1, all with both EOFs. Original closure covered 120 sealed
wait records and 124 absent PIDs/groups. At release, all 443 original input
pins (including built module, host, route and 15 unchanged media files) and
13 foreign WIP file hashes were unchanged, and both locks were free.
Runtime release SHA-256:
c1259a400ab9483dfaf4e59491ed6058280aa27c5da71a4e265b191b9fe7d87d.
Q3layout916770 is next; only this report remained owned after release.
Evidence retains exact Git SHA plus relevant binary patches, pins, compact logs/direct waits and four personally reviewed result-named PNGs: current untouched P1, native Quit panel, native confirmation, and the explicitly historical public completion/password screen. Unselected captures were removed; no source/media/WASM/frame archive was made. No box was created: ID null, state NOT_CREATED, expiry null, stop receipt NOT_APPLICABLE.
Next work requires a fresh original authorization and a way to complete personal gameplay review and dispatch inside the existing margin. This task does not authorize a retry, wider review interval, automatic rebinding or guessing the password. The authorized single original is consumed and closed.