[WIP] SDL2 thread proxying fixes #127
Conversation
|
This isn't actually WIP, but it does require emscripten-core/emscripten#9336 to land before merging. |
There was a problem hiding this comment.
The two that use the new APIs are definitely a nice cleanup. Might want to change the identical commit messages on the two mouse ones ("cursor creation"/"cursor setting"?), that confused me for a bit.
Not sure if there's a reason this doesn't just use MAIN_THREAD_EM_ASM* everywhere? (Haven't done anything with threads yet.)
| @@ -209,7 +233,7 @@ Emscripten_ShowCursor(SDL_Cursor* cursor) | |||
| curdata = (Emscripten_CursorData *) cursor->driverdata; | |||
|
|
|||
| if(curdata->system_cursor) { | |||
| EM_ASM_INT({ | |||
| MAIN_THREAD_EM_ASM_INT({ | |||
There was a problem hiding this comment.
Switch this to MAIN_THREAD_EM_ASM and drop the return. (I think EM_ASM was less flexible when this was written.)
| hot_y, | ||
| conv_surf->pixels | ||
| ); | ||
| } |
There was a problem hiding this comment.
Slightly annoying that we have to proxy this as it just wants to create an image from some data... But I don't think it can be avoided.
|
You might want to add |
This PR uses new APIs added in emscripten-core/emscripten#9336 to improve compatibility with
USE_PTHREADS=1.