Handling of DDB in tests?

[Wolfram Sang]

Hi,

while trying to deliver a testcase for Bug #14891 (Bitmaps with offset
bits do not display correctly) I tried this patch:

https://testbot.winehq.org/JobDetails.pl?Key=1740

Two VMs reported the correct pixel as 0x0000ffff instead of 0x00ffffff.
(The truly failed pixel would have 0x00010101 or similar). I'd assume that
is because they use 16 bit display-depth? If so, is it "good enough" to
check for both of these values as expected results? I am a bit confused
because LoadImage creates a DDB (via CreateCompatibleBitmap()) and the
pixeldata is thus allowed to look like anything. Still, there are checks
with GetPixelData and memcmp in the other tests (gdi32/tests/bitmap.c).
But maybe I am wrong or missing something?

Thanks,

Wolfram