On 09/16/11 15:42, Francois Gouget wrote:
On Fri, 16 Sep 2011, Jacek Caban wrote: [...]
hres = CoCreateInstance(&CLSID_WebBrowser, NULL,CLSCTX_INPROC_SERVER|CLSCTX_INPROC_HANDLER, &IID_IUnknown, (void**)unk);
- ok_(__FILE__,line)(hres == S_OK, "Creating WebBrowser object failed:
%08x\n", hres);
- if (hres == REGDB_E_CLASSNOTREG)
win_skip_(__FILE__,line)("WebBrowser is not supported\n");[...]
The failure is valid here, it's the testbot that is not smart enough. The tested DLL is not present on WINNT, so the test should be skipped automatically, like winetest does.
Yes, that's why I used a wine_skip() rather than a skip(). I think this approach is valid too. My understanding is that in theory CLSID_WebBrowser could be provided by any other dll. So given that this test makes no direct use of ieframes.dll (that is does not link with it nor loads it at run time), that makes checking for the presence of ieframes.dll a bit questionable.
IEs older than 7 had WebBrowser control implemented in shdocvw.dll. We surely don't want to run tests on anything older than IE6. IE6 results very little value in this case, so I consider DLL presence check as a nice way of skipping tests on too old platforms and I've even stripped the code from no longer needed checks that you want to add when I was moving them from shdocvw to ieframe, In this case, skipping tests is really a job for testbot.
Jacek