CNA-VGAP-032: Input behaviour on real hardware is unverified: the 15-task manual hardware checklist is blocked, and the focus-loss keyboard test checks only the bridge's accumulator

CNA snapshot 009d40f5  ·  Known Issues › Verification gaps  ·  source links pinned to 009d40f5

Controllers, touch, IME, non-US layouts, joysticks and host sensors are recorded as unverified on hardware, and the focus-loss keyboard test reads InputManager's legacy state through a local shim, not the public Keyboard::GetState.

Identifier

CNA-VGAP-032

Category

Verification gap

Subsystem

Testing & evidence

Status

Open

Verified against

CNA 009d40f5 (009d40f5dd085c4e674d3479675fac84b12b3e0a)

Evidence basis

Source-verified: read at TARGET, not executed

Tests touching this area

Yes: see Current tests

Affected contract

Keyboard::GetState across focus loss; GamePad, TouchPanel, text input/IME and device input on real hardware

Expected behaviour

Input semantics that depend on devices and window systems need evidence from those devices and window systems.

Actual behaviour at TARGET

plan_input.md Phase 11 (P11-001..015, the manual hardware checklist) is open, every task marked blocked, and the hardware matrix in input-manual-verification-results.md has every controller, touchscreen, IME and Czech-layout cell unchecked. WindowFocusLostDoesNotClearHeldKeysMatchingFna in SdlInputBridgeKeyboardTests.cpp reads a file-local Keyboard struct whose GetState is InputManager::GetKeyboardState(); the public Keyboard::GetState reads the platform keyboard snapshot (IPlatformKeyboard), and no test reads it across focus loss. Platform-level focus-loss release is pinned only for Wayland (WaylandProtocol.KeysHeldAtEnterAreStateNotEventsAndLeaveReleasesThemQuietly, in a suite no workflow runs) and, at event-mapper level, for Win32 (Win32EventMapping.TracksHeldKeysAndReleasesThemOnFocusLoss); X11 releases all keys on FocusOut and SDL3 reads SDL's keyboard state, with no focus-loss test for either.

Source locations

• plans/plan_input.md — Phase 11 manual hardware validation: 15 tasks
• docs/input-manual-verification-results.md — hardware verification matrix, all cells unchecked
• modules/input/tests/CNA/Internal/Input/SdlInputBridgeKeyboardTests.cpp — local Keyboard shim over InputManager

Evidence

Checked by reading at 009d40f5; not executed.

Independent re-verification: Checked by reading at 009d40f5; not executed. Platform focus-loss tests found by searching modules/platform/tests and modules/input/tests for FocusLost, FocusOut and KILLFOCUS.

Focused reproduction

No focused reproduction is known. Nothing has been invented here; the evidence above is what exists.

Current tests

Headless unit tests cover the input state machine with canned platforms and synthetic events.

Regression test

Per-platform focus-loss tests through the public Keyboard::GetState; recorded hardware runs using the template in the results document.

Blast radius

Evidence only; controller mappings, rumble, touch gestures, IME composition and layouts may differ on real devices.

Workaround

No workaround is known.