On Tue Mar 5 20:50:52 2024 +0000, Jeffrey Smith wrote:
> As the documentation states that the updated region _may_ be larger than
> the specified region, this seemed safer.
Yeah, but I don't really want to trust that programs are following the letter of the documentation either, especially with an API this old. I'd rather at least test exact values, even if we have to mark them todo.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5161#note_63598
On Tue Mar 5 19:10:38 2024 +0000, Zebediah Figura wrote:
> Any reason not to test the exact values?
As the documentation states that the updated region _may_ be larger than the specified region, this seemed safer.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5161#note_63597
On Tue Mar 5 19:10:37 2024 +0000, Zebediah Figura wrote:
> Can there be multiple updates? You mention the possibility of "at least"
> one update in your tests; should we be storing an array instead?
Multiple calls to ForceUpdate seem entirely reasonable.
This was part of trying to keep things simple for the implementation and effectively have a single dirty-rectange.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5161#note_63596
On Tue Mar 5 19:10:31 2024 +0000, Zebediah Figura wrote:
> Where do these fail?
- The first fails on the testbot (XP/2003) with result DDERR_CANTLOCKSURFACE.
- The second fails on a local box on Win10 (with a native d3drm.dll copied in) - the color returned from get_surface_color is always 0x00ff00ff.
FWIW, the `test_update_surfaces*` tests _do_ run properly on a local box of mine, running XP bare-metal.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/5161#note_63595