1/3 is unfortunately causing some (strange) test failures here:
00f0:fixme:quartz:mpeg_splitter_filter_init_gst unexpected format 5
00f0:fixme:quartz:mpeg_splitter_filter_init_gst unexpected format 1
amstream.c:512: Test failed: Got hr 0.
I see this on the CI too.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5430#note_67851
--
v2: ntdll: Store exception reporting flags for debug events.
ntdll: Store exception reporting flags on suspend.
ntdll: Store exception reporting flags in server context.
ntdll: Set exception reporting flags in NtGetContextThread().
ntdll/tests: Add tests for CONTEXT_EXCEPTION_REQUEST.
ntdll/tests: Work around exception test failures on Wine WoW.
https://gitlab.winehq.org/wine/wine/-/merge_requests/5480
Thanks for the review.
> They fail on my Windows XP box, r250 GPU, see the attached log file. It seems that It doesn't follow the alpha = 0xff behavior of newer GPUs and just returns alpha 0x0 in either case. The window positioning test needlessly depends on it. If I mask out the alpha channel in test_get_front_buffer_data_windowed_positioning() it passes.
Interesting. I'd be perfectly fine with dropping the separate alpha test entirely. I also meant to mask out the alpha but I guess I forgot it there or lost it during rebase/cleanup.
> if (foo) { something(); }
> Please put the { on a new line consistently.
Ah, I'm trying. I thought I got it right but I guess old habits die hard. I use the same line { in every other project, so not surprised that I slipped up a couple of times. I'll fix that.
> test_swapchain_buffer_swapping() and test_get_front_buffer_data_alpha() leak your initial device, the one you use to populate present_parameters
> In the case of test_swapchain_buffer_swapping I don't see why you need to call IDirect3D9::CreateDevice yourself - what prevents you from using the device returned by create_device?
Oops, that's a leftover of my original approach. I originally meant to create a new device for each test case and pass the required presentation params but later changed it to just use a single device created using `create_device` and reset that as needed. I meant to get rid of the second device that's created manually.
I will also look into using the current monitor resolution for the fullscreen tests.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5500#note_67838