Re: [PATCH v9 0/10] MR4982: Load and play MIDI

[Rémi Bernon (＠rbernon)]

Rémi Bernon (@rbernon) commented about dlls/dmime/tests/dmime.c:
> +{
> - IDirectMusicSegment8 *segment = NULL;
> - IDirectMusicLoader8 *loader;
> - WCHAR test_mid[MAX_PATH];
> - HRESULT hr;
> -
> - load_resource(L"test.mid", test_mid);
> -
> - hr = CoCreateInstance(&CLSID_DirectMusicLoader, NULL, CLSCTX_INPROC_SERVER,
> -
>        &IID_IDirectMusicLoader8, (void **)&loader);
> - ok(hr == S_OK, "got %#lx\n", hr);
> - hr = IDirectMusicLoader8_LoadObjectFromFile(loader, &CLSID_DirectMusicSegment,
> -
>        &IID_IDirectMusicSegment, test_mid, (void **)&segment);
> - todo_wine ok(hr == S_OK, "got %#lx\n", hr);
> - IDirectMusicLoader8_Release(loader);
> - if (segment) IDirectMusicSegment8_Release(segment);

With this commit alone we return S_OK and the test doesn't seem to need a todo_wine (or the `if (segment)` below).

I guess this is probably because we eat the error somewhere in the call chain, but you should adjust the test results so that there's no failure (or spurious success) regardless of which commit we run them at.

You can keep the test as it is, or you can also start by fixing wherever the error is swallowed, so that the todo_wine is indeed needed.