http://bugs.winehq.org/show_bug.cgi?id=22744
--- Comment #2 from Wylda wylda@volny.cz 2010-09-12 04:53:01 CDT --- Created an attachment (id=30708) --> (http://bugs.winehq.org/attachment.cgi?id=30708) Screen shot showing the problem
According to Appdb the game rendered textures correctly around 0.9.15.
Good catch. I normally don't watch attached screen shots at AppDB ;)
As i don't like regression, i gave this one a try. Probably hardest regression test ever made...
So some textures are missing and also notice from attached screen shot, that graphics is also distorted (missile locked airplane, unreadable font).
1. I did several regression tests between 0.9.15 and 0.9.41. Particular commit(wine-0.9.36-240-gd9fef10):
commit d9fef10b71b9add27cfaeffa6d7e3ab42342bba7 Author: Stefan Dösinger stefan@codeweavers.com Date: Wed May 9 11:49:27 2007 +0200
ddraw: Some caps fixes.
Cards at DirectX7 times did not support non power of 2 textures, and some games check the capatiblity flags incorrectly. Rollcage expects the D3DPTEXTURECAPS_POW2 (limitation) flag set and fails with the reference rasterizer which support NP2 textures and doesn't have the cap set.
2. There are another bug reports suffering from this commit, see bug 9665, bug 10133, bug 10164, bug 21878.
3. Revert of this patch after git checkout makes that problem go away (can't be tested directly). Attached hack proves correctness of regressions test.
4. Adding author of this patch to CC.
--private keyword: bisected