Support EGL_KHR_display_reference to clean up EGL properly
#1588
Comments
|
cc @i509VCB given that you had some concerns about not calling |
|
This is definitely a solution to the problem for displays which glutin has created and owns. For glutin do we want to always set that attribute if supported? What about things which borrow the display, should cloning the display increment the ref count of the display reference or should we continue with the interior mutability in that case? For displays which glutin does not own or takes ownership of it appears we can query if Also what how does driver support look for this extension? |
|
it's not a driver extension, it's a platform extension, like a We want to always set it if supported and don't anything otherwise. We should increment on The |
This resolves a past issue where glutin would not be able to terminate a display due to risk of two displays being created from the same native display. Besides not terminating EGL raises issue with nvidia on Wayland, thus the `api::egl::Display::terminate` is added to help clients call this function when calling `eglTerminate` is generally unsafe to do automatically when dropping the `Display`. Links: alacritty/alacritty#7146 Fixes: rust-windowing#1588
This resolves a past issue where glutin would not be able to terminate a display due to risk of two displays being created from the same native display. Besides not terminating EGL raises issue with nvidia on Wayland, thus the `api::egl::Display::terminate` is added to help clients call this function when calling `eglTerminate` is generally unsafe to do automatically when dropping the `Display`. Links: alacritty/alacritty#7146 Fixes: rust-windowing#1588
This resolves a past issue where glutin would not be able to terminate a display due to risk of two displays being created from the same native display. Besides not terminating EGL raises issue with nvidia on Wayland, thus the `api::egl::Display::terminate` is added to help clients call this function when calling `eglTerminate` is generally unsafe to do automatically when dropping the `Display`. Links: alacritty/alacritty#7146 Fixes: rust-windowing#1588
This resolves a past issue where glutin would not be able to terminate a display due to risk of two displays being created from the same native display. Besides not terminating EGL raises issue with nvidia on Wayland, thus the `api::egl::Display::terminate` is added to help clients call this function when calling `eglTerminate` is generally unsafe to do automatically when dropping the `Display`. Links: alacritty/alacritty#7146 Fixes: rust-windowing#1588
This resolves a past issue where glutin would not be able to terminate a display due to risk of two displays being created from the same native display. Besides not terminating EGL raises issue with nvidia on Wayland, thus the `api::egl::Display::terminate` is added to help clients call this function when calling `eglTerminate` is generally unsafe to do automatically when dropping the `Display`. Links: alacritty/alacritty#7146 Fixes: rust-windowing#1588
This extension defines a ref counting for the
EGLDisplaywhich we could use to properly calleglTerminateinstead of leaking like we have it now.The text was updated successfully, but these errors were encountered: