Re: [PATCH v3 0/4] MR7928: coremessaging: Partially implement CreateDispatcherQueueController().

[Rémi Bernon (＠rbernon)]

Rémi Bernon (@rbernon) commented about dlls/coremessaging/tests/coremessaging.c:
> - hr = IAsyncAction_QueryInterface( operation, &IID_IAsyncInfo, (void **)&async_info );
> - ok_(__FILE__, line)( hr == S_OK, "got IAsyncAction_QueryInterface hr %#lx.\n", hr );
> -
> - hr = IAsyncInfo_get_Status( async_info, &status );
> - ok_(__FILE__, line)( hr == S_OK, "got IAsyncInfo_get_Status hr %#lx.\n", hr );
> - ok_(__FILE__, line)( status == Started, "got IAsyncInfo_get_Status status %d.\n", status );
> - while (PeekMessageW( &msg, NULL, 0, 0, PM_REMOVE ))
> - {
> -
>    TranslateMessage( &msg );
> -
>    DispatchMessageW( &msg );
> - }
> - ret = WaitForSingleObject( event_handler->event, INFINITE );
> - ok_(__FILE__, line)( !ret, "Unexpected wait result %lu.\n", ret );
> - hr = IAsyncInfo_get_Status( async_info, &status );
> - ok_(__FILE__, line)( hr == S_OK, "got IAsyncInfo_get_Status hr %#lx.\n", hr );
> - flaky ok_(__FILE__, line)( status == Completed, "got IAsyncInfo_get_Status status %d.\n", status );

I don't think we want to introduce new flaky tests, it's meant to workaround existing tests that are difficult to stabilize but it's always preferable to either make the test pass reliably, or, maybe, not test the behavior if it's too difficult.