LTris normal manual game and native Quit, 2026-11-17
The line-clear goal failed. One normal original game reached Game Over after 13 pieces with Lines 0 and Score 0. Original native Pause/resume and horizontal piece movement were observed. A visibly selected native Quit was activated by Enter, consumed at the native-input-return boundary, independently physically released with ACK, and followed by native exit 0. This is neither a line clear nor a score consequence.
Evidence: scratch/runs/20261117-ltris-original-manual-line-clear-onward/result.json.
Sixteen selected named whole frames are retained; redundant images were pruned
with their hashes and capture metadata preserved. No OCR, solver, symbolic board
memory, imported moves, configuration/save/clock/state changes or second game
was used. The default New Game screen showed Normal mode, level 0 and preview on.
Player 1 controls and global Pause P were personally reviewed in the original.
The authored SDL fixture first completed the personally reviewed full route: black menu, green board, red Pause, green resume. Pause and resume reached native consumption at capture ages 4.4916 and 3.5701 seconds. Both independent UPs were acknowledged. Preparing a persistent operator before capture removed repeated helper imports from the review-to-action interval; each action was still chosen only after whole-image review. The five-second gameplay bound, 500 ms guarded transport/ACK, 1750 ms independent UP and 3000 ms capture bound were unchanged. Seven expired local intents across fixture and original remained NOT_SENT. There was no actual runtime input/capture UNKNOWN on either working endpoint.
Preparation failures remain separate. An initial fixture paired the retained module with a newer source/memory-map mirror and exited 2 before a control endpoint. A preliminary original launcher then refused missing sealed candidate metadata before starting the guest; no board, binding or input occurred there. The small existing, hash-checked media metadata was added to the transient inventory, and the authorized gameplay sequence subsequently ran. Neither failure is evidence of a game input defect.
The working runtime reused matching published source b29ef843aee74fe366aac3698704e3238797ed9e
and retained module SHA-256
533c599a144837890b6b45f6cbace2e52e9b4f48a3a8dd08b4261cd20223b1a1,
with the published inventory resolver from ad27fc08f. Its owner had reported
focused resolution/startup/controller tests and a locked canonical build passing.
Those owned files were not edited. Actual compilation/Worker module receipts,
source inventories, relevant binary diffs and the existing compiled-source proof
are retained or hash-referenced. All 43 media references matched before and
after. This was not a test of an unqualified moving main module.
Native and independent watchdog actual waits have both EOFs; the watchdogs did not fire. The working fixture was explicitly stopped after its completed route. Its enclosing launcher and the original gameplay enclosing launcher returned 143 without writing their actual parent-wait records. That parent-wait/EOF gap is explicitly UNKNOWN and cannot be reconstructed from absent PIDs or the native wait. All named native/watchdog/supervisor PIDs are absent and listeners refuse connections. Native Quit does not establish complete enclosing lifecycle proof.
No public runtime defect was proved, so there is no speculative runtime repair, new build claim or affected-original retry. Only this report is committed. Home free space remained above 1.5 GB at recorded checks. No workers, boxes, source archives, media copies or deployment were created; foreign WIP and held investigations were preserved. Box ID/expiry are null, state NOT_CREATED and stop receipt NOT_APPLICABLE.
The gameplay allowance was used. A further original attempt needs renewed authorization, and its enclosing launcher must durably retain actual parent waits and both EOFs. The line-clear/score criterion remains open. Release/stop.