https://bugs.winehq.org/show_bug.cgi?id=41930
--- Comment #44 from Bryan Varner bryan@varnernet.com --- This is where I get to confess I was wrong about the SRCCOPY being the wrong thing.
I hacked up a quick and dirty StretchBlt that did the introspection on the src, and swapped the ROP to the SRCPAINT....
The _initial_ paint when the settler is active on the map (where it paints the translucent diamond pattern into the offscreen mesa context then blts back to the main screen) worked. However, the map layer never got updated properly again. Every draw after was black.
So, the SRCPAINT vs SRCCOPY and premultiplied alpha idea I've been harboring for a year or so is _not_ the issue. I've been wrong about that for a very long time.
However, in reviewing the ticket, and looking at the results of what was happening when I did the SRCPAINT hack, it has me wondering if I wasn't onto something at the tail end of Comment 30.