Re: [mscoree/tests 4/4] Fix test failures on W2K and earlier

[Vincent Povirk]

I think this test can be removed.

I wanted to verify that LoadLibraryShim would succeed for a dll that
is absent from the .NET system directory but present in system32. I
had heard of LoadLibraryShim succeeding for gdiplus, so I was very
surprised when this failed.

I was so surprised that I did the same test with gdi32, because I was
sure that would never be duplicated. (And it looks like it isn't.)

I think that the older versions must have gdiplus in the .NET system
directory so that the System.Drawing libraries (which are a thin
wrapper around gdiplus) can load it. I guess that .NET searches for
libraries in the .NET system directory and the regular path, meaning
this is not needed on later versions of Windows that have gdiplus in
system32.

Anyway, I think this shows that the success or failure is not
important in this case, and we don't need the test.

On Mon, Nov 22, 2010 at 1:37 PM, Paul Vriens <paul.vriens.wine(a)gmail.com> wrote:
> Changelog
>  Fix test failures on W2K and earlier
>
> --
> Cheers,
>
> Paul.

[mscoree/tests 4/4] Fix test failures on W2K and earlier

[Paul Vriens]

On 11/22/2010 05:57 PM, Vincent Povirk wrote:
> I think this test can be removed.
>
> I wanted to verify that LoadLibraryShim would succeed for a dll that
> is absent from the .NET system directory but present in system32. I
> had heard of LoadLibraryShim succeeding for gdiplus, so I was very
> surprised when this failed.
>
> I was so surprised that I did the same test with gdi32, because I was
> sure that would never be duplicated. (And it looks like it isn't.)
>
> I think that the older versions must have gdiplus in the .NET system
> directory so that the System.Drawing libraries (which are a thin
> wrapper around gdiplus) can load it. I guess that .NET searches for
> libraries in the .NET system directory and the regular path, meaning
> this is not needed on later versions of Windows that have gdiplus in
> system32.
>
> Anyway, I think this shows that the success or failure is not
> important in this case, and we don't need the test.

I was smart enough to make that the last patch (after briefly talking to 
you in Paris). I'll change the patch and resend the series.

-- 
Cheers,

Paul.