[4/4] (Try4) msi/tests: Fix package test when run on root drive directory.

[Nicolas Le Cam]

2009/4/13 James McKenzie <jjmckenzie51 at earthlink.net>:
> Nicolas Le Cam wrote:
>> 2009/4/13 James McKenzie <jjmckenzie51 at earthlink.net>:
>>
>>> Nicolas Le Cam wrote:
>>>
>>>> James,
>>>>
>>>> Here are updated patches for part 3 & 4 of my previous patch set. Tell
>>>> me if you think I could submit them to wine-patches.
>>>>
>>>> For patch 3, the only todo_wine is when launched from root drive dir
>>>> as explained in my previous mail.
>>>> For patch 4, I skip a test if test is executed on root drive dir, as
>>>> it doesn't make sense to execute it in this case.
>>>>
>>>> If needed I can remove the skip and change expected value to be first
>>>> fixed drive when test is run on root drive dir. But this is already
>>>> tested by patch 3 (when run on a root drive dir).
>>>>
>>>>
>>>>
>>> Did these work with WindowsXP or Windows Vista as expected?  I don't
>>> have them here to test, only Wine.
>>>
>>> I would defer the todo_wine to others with more experience, as this
>>> marks the tested function needs more work with Wine and my experience
>>> with msi is only as a user.
>>>
>>> James McKenzie
>>>
>>>
>>>
>>>
>>
>> It worked as expected on Win2k and Wine. I can test it on WinXP if
>> wanted, but I don't own any Vista.
>>
>
> I would try it on XP as that is the default Windows we try to reproduce now.
>> The todo_wine is to handle root drive directories where Windows output
>> first fixed drive where Wine currently output an empty string as
>> demonstrated on .
>>
>>
> Yep, that I do know about.
>> Fix seems to be simple. We just need to break the for loop after the
>> GetDriveType test in ACTION_SearchDirectory function in appseach.c if
>> path is only three characters long (attr should also be checked
>> against INVALID_FILE_ATTRIBUTES just after it).
>>
> Simple fixes always turn out to be the most complex whenever I've dealt
> with them.
>> ACTION_CheckDirectory could perhaps be updated to handle root drive
>> directories too, but I'm not certain with this one.
>>
>>
> I agree.  I don't know of any software package that should be installed
> to any root drive that would be handled by msi.
>> As I don't have any experience with msi at all even as user, I will
>> certainly wait for someone to validate what I'm saying.
>>
> I do have experience using msi as the installer.  In my experience, no
> program installed by me was to a root directory.  However, some programs
> installed were run from the D:\ and C:\.  So, it may be a good idea to
> insure that Wine's msi can handle these cases.
>
>> Thanks for review
> You are welcome.  I hope that the root directory case can be corrected
> as well.
>
> James McKenzie
>
>

If test passes on WinXP, I will submit patches to wine-patches with a
fix for msi. If something is wrong I will surely get feedback as it
was the case for previous series.

This will only let rpcrt4 test as a failing test (crash) on my Win2k
test platform (others are caused by the VM). I will run all tests on a
root drive directory to see if I can find more failures, then switch
to XP testing.

-- 
Nicolas Le Cam