TuxRacer Practice mouse-consumption boundary

Historical source last updated, 2026-10-30. Runtime and implementation claims below describe that dated source; see current app notes and source for subsequent changes.

Exactly one original guest used the retained runner f96e9bb924700ee571e2904e097f80742ff55638399898ae1dc32454af05b02d and module 5a39225e1a8381ce1c4530d2ce7cd6311bd75576d10968c20e963baf033d41eb. No production adapter or game binary was changed. The controller was complete, identity-checked and smoke-tested before startup. Its observed MainMenu condition triggered native mousemove320,260, then mousedown and mouseup with intervening captures. Button hold was579ms. Original production calls each forwarded once through a read-only observer.

The renderer canvas and window client area were640x480 at origin0,0. All three events retained lParam0x1040140, encoding320,260. WM_MOUSEMOVE dequeued and DispatchMessageA-dispatched at batch10283; down at10385 and up at10460. Motion, a held-button interval and separate dispatch batches did not produce visible Practice progression. Inspected aftermove, buttondown and postclick-settled images all show MainMenu. Practice, race and gameplay are NOT established. SDL-event and menu-callback consumption were not directly instrumented, so wndproc dispatch is the last proved consumption boundary.

Pinned release-source queries explain the intended path: TuxRacer0.61 winsys.c forwards SDL button coordinates; ui_mgr.c flips y to480-y; game_type_select.c positions Practice within x170..470 and top-origin y239..279; button.c requires matching in-bounds down/up, without a focused/hover prerequisite. SDL1.1.8's Win32 handler ordinarily uses message coordinates. Its relative path instead supplies zero coordinates and SDL_PrivateMouseButton falls back to stored SDL mouse position. The observer saw host pointer320,240 before down, after the requested move320,260. This is compatible with a centering warp, but neither relative-mode state nor SDL's private position was measured. It is a remaining hypothesis, not an established adapter defect or permission to patch one. Release-source provenance does not establish binary-private state.

One TTL1800 fork bx_a766zh4p came from bx_rht9232f after its exact snapshot/stopped timestamps and retained archive hash were verified. An independent watchdog was armed before first API execution. Runtime ended before cutoff; selected evidence was retrieved, and delegated owner cleanup verified snapshot completion2026-10-30T12:38:21.096Z and actual stop2026-10-30T12:38:51.869Z. No Boat remains running. Shared root, main, Telegram, public deployment and unrelated ownership were untouched. Evidence is under test/tuxracer-practice-mouse-consumption-runtime-oct30; coordinator receives completion-receipt.json and messageboard-note.txt for append-only shared-board delivery.