On 29/03/18 08:46, Jacek Caban wrote:
Hi Alistair,
diff --git a/dlls/wmp/tests/oleobj.c b/dlls/wmp/tests/oleobj.c index e4c2294436..58ba6081b1 100644 --- a/dlls/wmp/tests/oleobj.c +++ b/dlls/wmp/tests/oleobj.c @@ -970,10 +970,13 @@ static void test_wmp_ifaces(IOleObject *oleobj) ok(0 == lstrcmpW(url, filename), "%s != %s\n", wine_dbgstr_w(url), wine_dbgstr_w(filename)); SysFreeString(url);
/* Keep a reference since we are using the same media object, stops a crash (w2003std, wvistau64) */
IWMPMedia_AddRef(media); SET_EXPECT(GetContainer); hres = IWMPPlayer4_put_currentMedia(player4, media); ok(hres == S_OK, "put_currentMedia failed: %08x\n", hres); todo_wine CHECK_CALLED_OR_BROKEN(GetContainer);
IWMPMedia_Release(media);
IWMPMedia_Release(media);
Note that we already hold reference to media interface, so this should not be needed. If the patch helps, it means that we have a problem with ref counting somewhere else or Windows implementation is somehow broken. Given that the problem is visible only on one platform, I suspect the later. Maybe we should skip WMP tests on this platform?
Thanks, Jacek
For what it's worth, it seems to me that it'd be more desirable to work around a Windows bug rather than skip the test entirely.