On 5 June 2016 at 20:45, Stefan Dösinger stefandoesinger@gmx.at wrote:
Other than being in the old place, this test doesn't initialize Surface1 and randomly crashes when SetDisplayMode fails.
That's likely all true, but those would be reasons for fixing the test instead of removing it. Do the previous two tests in the series cover all the functionality tested by this test?
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
Am 2016-06-07 um 16:18 schrieb Henri Verbeet:
That's likely all true, but those would be reasons for fixing the test instead of removing it. Do the previous two tests in the series cover all the functionality tested by this test?
In the past you rejected even typo fixes in the old files :-) .
Covering the functionality in the new tests was the intention, yeah. The color table verification already exists in test_palette_gdi. The two patches add the GetDC color verification and Blitting tests, although they test the blit result through locking the surface instead of reading it through GetDC. Because the DIB is and always has been based on the P8 data this doesn't make a difference.
I didn't take the lock - blit - lock - blit test along because we have many other tests that check our texture upload and download codepaths. Since we're doing P8 conversion only when showing the final front buffer P8 transfers aren't special in that regard any more.
On 7 June 2016 at 21:42, Stefan Dösinger stefandoesinger@gmail.com wrote:
Am 2016-06-07 um 16:18 schrieb Henri Verbeet:
That's likely all true, but those would be reasons for fixing the test instead of removing it. Do the previous two tests in the series cover all the functionality tested by this test?
In the past you rejected even typo fixes in the old files :-) .
Well, yeah, fixing it would involve rewriting it for ddraw[1247].c.
-
Henri Verbeet -
Stefan Dösinger