Re: user32: Add a test for GetClassInfo, make it pass under Wine.

[Paul Vriens]

On 07/29/2010 08:55 AM, Dmitry Timoshkov wrote:
> - SetLastError(0xdeadbeef);
> - ret = GetClassInfoEx(0, "static", NULL);

Why not the explicit A-function? I see there is already an existing call 
in the current tests, but still.
> - ok(!ret, "GetClassInfoEx() should fail\n");
> - ok(GetLastError() == ERROR_NOACCESS, "expected ERROR_NOACCESS, got %d\n", GetLastError());
> -
> - SetLastError(0xdeadbeef);
> - ret = GetClassInfoExW(0, staticW, NULL);
> - ok(!ret, "GetClassInfoExW() should fail\n");
> - ok(GetLastError() == ERROR_NOACCESS ||
> -
>   GetLastError() == ERROR_CALL_NOT_IMPLEMENTED, /* win9x */

Wouldn't broken() be more appropriate?

user32: Add a test for GetClassInfo, make it pass under Wine.

[Dmitry Timoshkov]

Paul Vriens  wrote:
> On 07/29/2010 08:55 AM, Dmitry Timoshkov wrote:
> > - SetLastError(0xdeadbeef);
> > - ret = GetClassInfoEx(0, "static", NULL);
>
> Why not the explicit A-function? I see there is already an existing call 
> in the current tests, but still.

An API without a suffix is A by default.
> > - ok(!ret, "GetClassInfoEx() should fail\n");
> > - ok(GetLastError() == ERROR_NOACCESS, "expected ERROR_NOACCESS, got %d\n", GetLastError());
> > -
> > - SetLastError(0xdeadbeef);
> > - ret = GetClassInfoExW(0, staticW, NULL);
> > - ok(!ret, "GetClassInfoExW() should fail\n");
> > - ok(GetLastError() == ERROR_NOACCESS ||
> > -
> >   GetLastError() == ERROR_CALL_NOT_IMPLEMENTED, /* win9x */
>
> Wouldn't broken() be more appropriate?

Probably, but I'd argue that returning ERROR_CALL_NOT_IMPLEMENTED differs
from returning actually wrong error code.