Proposed fixes to build with MinGW-w64 8.0.0 and JWasm #475
Conversation
|
|
There was a problem hiding this comment.
Mostly looks good. I noticed one thing that I didn't catch before which I think would be best to tweak, which I highlighted inline. I can change that or you can — it doesn't matter to me. I'm running this through our internal CI and I'll update when that passes or fails.
loader/loader.c
Outdated
| const ULONG flags = CM_GETIDLIST_FILTER_CLASS | CM_GETIDLIST_FILTER_PRESENT; | ||
| #else | ||
| const ULONG flags = CM_GET_DEVICE_INTERFACE_LIST_PRESENT; |
There was a problem hiding this comment.
This is pretty clearly clear intended for the CM_Get_Device_Interface_List functions whereas we're using this in the CM_Get_Device_ID_List functions so I don't think this is quite correct. Interestingly, the value for this is zero, which is equivalent to CM_GETIDLIST_FILTER_NONE so it probably works ok. But that does mean you're not filtering out devices that have been removed (which is half the point we added this code). I think the solution I'd go with is probably just to add the numerical value instead so the behavior doesn't change. So I'd change this to:
const ULONG flags = 0x300; // Equivalent to (CM_GETIDLIST_FILTER_CLASS | CM_GETIDLIST_FILTER_PRESENT) which aren't always defined on MinGWThere was a problem hiding this comment.
Yes, that makes more sense.
My attempt was to find the equivalent in MinGW-w64, but if it results in 0 that doesn't make sense.
Go ahead and change it if you like.
See http://winlibs.com for a release of such GCC compiler See issue #474
Taken from KhronosGroup/Vulkan-Loader#475. Supersedes and reverts godotengine#43119 since the upstream change removes the need for that custom define.
#474
See http://winlibs.com for a release of a GCC compiler with MinGW-w64 8.0.0.