Using dispatch_pending in a busy loop here isn't ideal, but poll_dispatch_pending doesn't seem to work?
If GDK tries tries to destroy a surface, an EGL_BAD_DISPLAY error is produced, which causes a segfault. With
https://github.com/Smithay/smithay/pull/771 it logs instead of segfaulting. This seems to happen as long as GpuManager::new() is called, even if it isn't used.
Copying the dmabuf into a Vec is not ideal, but GDK doesn't seem to provide a way to get the EGLContext behind a Gdk::GLContext.
This should be updated to have an optional crate feature that provides abstractions based on smithay-client-toolkit to simplify the code needed here. We can also use that for an Iced example then.
Captures all outputs using
export_dmabuf.Using
dispatch_pendingin a busy loop here isn't ideal, butpoll_dispatch_pendingdoesn't seem to work?If GDK tries tries to destroy a surface, an
EGL_BAD_DISPLAYerror is produced, which causes a segfault. With https://github.com/Smithay/smithay/pull/771 it logs instead of segfaulting. This seems to happen as long asGpuManager::new()is called, even if it isn't used.Copying the dmabuf into a
Vecis not ideal, but GDK doesn't seem to provide a way to get theEGLContextbehind aGdk::GLContext.This should be updated to have an optional crate feature that provides abstractions based on
smithay-client-toolkitto simplify the code needed here. We can also use that for an Iced example then.