Saulius Krasuckas wrote:
- On Mon, 15 Aug 2005, Jacek Caban wrote:
I've tried to get other tests working on win 98, without success. It's possible that it won't work this way on 9x,
It seems I've found one bug in test/htmldoc.c:
| if(activeobj) { | - IOleInPlaceActiveObject_GetWindow(activeobj, &hwnd); | + hres = IOleInPlaceActiveObject_GetWindow(activeobj, &hwnd); | ok(hres == S_OK, "GetWindow failed: %08lx\n", hres);
Right, it's a bug.
And we have a failure in an old version of test on winME:
| htmldoc.c:502: Test failed: GetWindow failed: 80004005
- [1] writes:
| | IOleInPlaceActiveObject::GetWindow | | Always fails and returns E_FAIL for a windowless | control. Since this is documented in the current | OLE2 specification, there should not be any problem | with existing containers and frames.
So something might be missing here, I thought.
OTOH, winME always returns E_FAIL {80004005} in this test [2], while win98 seems to return HRESULT_FROM_WIN32(ERROR_ALREADY_EXISTS) {800700b7} [3].
And I'd expect E_FAIL here as GetWindow fails if HTMLDocument is not active in-place (and it's not here because UIActivate failed). Other tests show that it's correct.
Maybe you give it a test-try on win98 as there is no way to do it by WT now. :-(
We shouldn't care too much about it as it happens only in scenario when UIActivate fails that won't happen in Wine (and in nt). I've tested different ways UIActivate can fail on xp and Wine fails the same way.
Jacek