shell32: Fix tests to apply to Vista+

[Nikolay Sivov]

On 4/23/2010 20:46, Paul Vriens wrote:
> On 04/23/2010 06:30 PM, André Hentschel wrote:
>> Juan Lang schrieb:
>>>> My point of view is that its not an error on windows, with this 
>>>> patch it passes in windows and gets a todo_wine, which it actually 
>>>> is (like you say).
>>>
>>> Any test that allows all possibilities, as your patch does on Windows,
>>> is worthless.  That implies that it should be investigated, and, if
>>> it's truly a meaningless test, then remove it.
>>> --Juan
>>
>> It still has two benefits applying the patch: it states its a todo on 
>> wine and it documents the old behavior of windows as broken.
>
> But who guarantees that the old behavior is broken and not the new one?
>
> I'm with Juan and Nikolay on this one, we need more investigation.
>
An update here for anyone who cares. I dumped returned pidls (results 
are here https://testbot.winehq.org/JobDetails.pl?Key=1725&log_101=1#k101).
Vista failure caused by SHParseDisplayName returning pidl with a root at 
UsersFiles (alias for user home directory in Users). Parsing from 
desktop returns pidl with root at My Computer.

Looks like SHParseDisplayName now uses this virtual folder (user home 
alias) as a root, that's all, and there's no way to switch to pre-vista 
behavior.