CNA-BUG-220: cna_game_destroy's comment says Shutdown has disposed the canonical graphics device; it has not

CNA snapshot 009d40f5  ·  Known Issues › Current bugs  ·  source links pinned to 009d40f5

A comment in cna_game_destroy claims CGame::Shutdown already disposed the game's canonical GraphicsDevice so C subscribers saw its Disposing event, but Game::Dispose(true) never disposes the Game-owned device.

Identifier

CNA-BUG-220

Category

Bug

Subsystem

Documentation & release tooling

Status

Open

Verified against

CNA 009d40f5 (009d40f5dd085c4e674d3479675fac84b12b3e0a)

Severity

Low (a triage suggestion, not a project priority)

Evidence basis

Source-verified: read at TARGET, not executed

Tests touching this area

Yes: see Current tests

Affected contract

cna_game_destroy teardown order (maintainer documentation in CnaCApiRuntime.cpp)

Expected behaviour

The comment that explains why the C-side device adapter state may be dropped states the real order: the canonical device is disposed only when the game object is destroyed, after the handle is released.

Actual behaviour at TARGET

After game->Shutdown() the comment reads “Shutdown has already disposed the canonical graphics device, so any C subscriber has observed its Disposing event”. CGame::Shutdown calls Game::Dispose(), and Game::Dispose(true) disposes components, the content manager and the registered IGraphicsDeviceService; GraphicsDeviceManager::Dispose(true) raises its own DeviceDisposing but never disposes a Game-owned device. The canonical GraphicsDevice raises Disposing from its destructor, after the adapter state has already been reset. The comment on ResetGraphicsDeviceAdapterState in modules/c-api/src/CnaCApiRuntimeDetail.hpp repeats the premise (the game-destroy path calls it "once the canonical device has already raised its own Disposing event").

Source locations

• modules/c-api/src/CnaCApiRuntime.cpp — cna_game_destroy comment after the Shutdown call; CGame::Shutdown
• modules/c-api/src/CnaCApiRuntimeDetail.hpp — the comment on ResetGraphicsDeviceAdapterState repeats the premise
• modules/runtime/src/Game.cpp — Game::Dispose(bool) does not dispose GraphicsDevice_
• modules/runtime/src/GraphicsDeviceManager.cpp — GraphicsDeviceManager::Dispose(bool): delete only when ownsGraphicsDevice_
• modules/c-api/include/CNA/C/graphics_device.h — device event registrations must be released before cna_game_destroy succeeds

Evidence

Checked by reading CnaCApiRuntime.cpp, Game.cpp and GraphicsDeviceManager.cpp at 009d40f5. The C API was not built or run. Because graphics_device.h requires device event registrations to be released before cna_game_destroy succeeds, no live C subscriber is known to be affected; whether a C caller can ever observe the canonical device's Disposing event was not established.

Focused reproduction

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

Current tests

The C API runtime tests (not built in CI) exercise cna_game_destroy; none asserts on the canonical device's Disposing event.

Regression test

Correct the comment; if the C API is meant to deliver the canonical device's Disposing event, add a C test that subscribes, destroys the game and expects exactly one callback.

Blast radius

Maintainers of the C API teardown path; no behaviour is known to depend on the false statement.

Workaround

No workaround is known.