Re: MSHMTL: Added QueryStatus implementation

[Jacek Caban]

Saulius Krasuckas wrote:
> - On Sun, 14 Aug 2005, Jacek Caban wrote:
> > I'm not sure if it is a correct fix. Although it avoids crashes,
> > much less functionality of QueryStatus is tested. Before
> > your patch  all possible commands were tested.
>
> This is true only for nt family windows. :-P
> On 9x test function exits on an exception.

As I wrote before, it'll be probably true for 9x family after
UIActivate succeeded.
> > it seems to be a case similar to crashes on xp while HTMLDocument is not 
> > active in place. In this case it is not active in place because 
> > UIActivate fails. Looking at the test results, UIActivate is a problem 
> > on every win 9x and causes most tests to fail. So first UIActivate 
> > should be fixed and then, if this problem remains, we'll be sure that 
> > your fix is correct.
>
> I don't understand much in all this OLE/In-Place stuph, so I agree, but 
> still I doubt this would be a sufficient to proof my patch is correct. 
> :-]
> > Presently if you want to make this test not to crash, you may disable 
> > IOleCommand tests if UIActivate fails.
>
> I will try, but what should be done with this (inside test_HTMLDocument)?
>
> |         OLECMD cmd[2] = {
> |             {OLECMDID_OPEN, 0xf0f0},
> |             {OLECMDID_GETPRINTTEMPLATE+1, 0xf0f0}
> |         };
> | 
> |         hres = IOleCommandTarget_QueryStatus(cmdtrg, NULL, 2, cmd, NULL);
> |         ok(hres == OLECMDERR_E_NOTSUPPORTED,
> |                 "QueryStatus failed: %08lx, expected OLECMDERR_E_NOTSUPPORTED\n", hres);

It should be disabled as well.
> Or IOW, why first call generates exception on OLECMDID_OPEN command while 
> second do not?
>
> |         hres = IOleCommandTarget_QueryStatus(cmdtrg, NULL, 2, cmd, NULL);
> |         hres = IOleCommandTarget_QueryStatus(cmdtrg, &IID_IHTMLDocument2, 2, cmd, NULL);
>
> OLECMDID_OPEN seems to be supported at least on winME.  I am curious, why?

The second call returns after checking that IID_IHTMLDocument2 is not a 
valid command
group id and it doesn't have to check anything else like the current 
state of
HTMLDocument.
> > My guess is that it requires to be loaded before a call to UIActivate,
>
> What do you exactly mean by "it"?

HTMLDocument
> > so the attached patch might help (even if so, a few other tests will 
> > fail, but currently the most interesting is UIActivate),
>
> No, the patch didn't help.

OK, I'll take a look at this today or tomorrow.
> Is something like this OK? I have "borrowed" boolean variable
> expect_SetActiveObject_active.

Yes, it looks good.

Thanks,
    Jacek