Failure in gdi/tests/metafile.c

[Uwe Bonnes]

Hallo,

../../../tools/runtest -q -P wine -M gdi32.dll -T ../../.. -p gdi32_test.exe.so metafile.c
failes like:

metafile.c:457: Test failed: (0,0)->(10,10), expected (0,0)->(18,67)
metafile.c:457: Test failed: (0,0)->(10,10), expected (0,0)->(18,67)

Some discussion on #winehackers brought up some clues, but no real fix.

Shouls the test be a todo_wine?

[Felix Nawothnig]

Uwe Bonnes wrote:
> ../../../tools/runtest -q -P wine -M gdi32.dll -T ../../.. -p gdi32_test.exe.so metafile.c
> failes like:
>
> metafile.c:457: Test failed: (0,0)->(10,10), expected (0,0)->(18,67)
> metafile.c:457: Test failed: (0,0)->(10,10), expected (0,0)->(18,67)
>
> Some discussion on #winehackers brought up some clues, but no real fix.
>
> Shouls the test be a todo_wine?

If the test is failing since it was added: yes.
If it's a regression: no.

But since this test is failing for a *long* time this is hard to figure 
out...

Felix

[Robert Shearman]

Uwe Bonnes wrote:
> Hallo,
>
> ../../../tools/runtest -q -P wine -M gdi32.dll -T ../../.. -p gdi32_test.exe.so metafile.c
> failes like:
>
> metafile.c:457: Test failed: (0,0)->(10,10), expected (0,0)->(18,67)
> metafile.c:457: Test failed: (0,0)->(10,10), expected (0,0)->(18,67)
>
> Some discussion on #winehackers brought up some clues, but no real fix.

I wrote the test, but I'm beginning to wish I hadn't. The failures are 
either one of two things: differences in when the transforms are updated 
in Win9x mode or possibly because it is too sensitive to the DPI of the 
screen. It definitely shouldn't be a todo_wine; the test should either 
be fixed if possible or deleted.

[Felix Nawothnig]

Robert Shearman wrote:
> I wrote the test, but I'm beginning to wish I hadn't. The failures are 
> either one of two things: differences in when the transforms are updated 
> in Win9x mode or possibly because it is too sensitive to the DPI of the 
> screen.

It's definitly not caused by differing DPIs anymore, I fixed that some 
weeks ago by using much smaller numbers.

A summary what is right/wrong:

For MM_ANISOTROPIC both wine and the test is correct.
The MM_TEXT test is correct, wine is broken when emulating Win9x.
The NULL test is wrong on Win9x.

Felix