Am 08.01.2019 um 01:42 schrieb Daniel Ansorregui <mailszeros@gmail.com>:

Signed-off-by: Daniel Ansorregui <mailszeros@gmail.com>
---
include/d3d9types.h | 2 ++
1 file changed, 2 insertions(+)

diff --git a/include/d3d9types.h b/include/d3d9types.h
index 4891858b3d..2e5648a8f6 100644
--- a/include/d3d9types.h
+++ b/include/d3d9types.h
@@ -827,6 +827,8 @@ typedef enum _D3DFORMAT {
    D3DFMT_MULTI2_ARGB8         = MAKEFOURCC('M', 'E', 'T', '1'),
    D3DFMT_G8R8_G8B8            = MAKEFOURCC('G', 'R', 'G', 'B'),
    D3DFMT_R8G8_B8G8            = MAKEFOURCC('R', 'G', 'B', 'G'),
+    D3DFMT_DF16                 = MAKEFOURCC('D', 'F', '1', '6'),
+    D3DFMT_DF24                 = MAKEFOURCC('D', 'F', '2', '4'),

The tests and implementation should in theory be able to compile with Microsoft's and Mingw's headers, although nobody tests that (not even my hacky visual studio setup), and there are likely other typos that currently prevent that. For the tests just put MAKEFOURCC('D', 'F', '1', '6') etc in the place where a D3DFMT is expected.

(Every now and then people try to compile Windows programs against our headers and libraries, e.g. to create a Wine port of an application to a non-x86 platform. See https://wiki.winehq.org/Winelib for details. That's the reason why we're aiming for header compatibility)