This helper would be responsible for:

1. Create the renderer and allocator for a backend
2. For each output, select a suitable format and create a swapchain for the primary buffers and the cursor buffers
3. On modeset, allocate a blank buffer if none is provided
4. Fallback to modifier-less buffers on modeset failure

All of this previously lived in the backends, and was recently lifted up to wlr_output. Extracting it in a helper would allow to simplify wlr_backend/wlr_output and make it un-opinionated (stop making decisions like picking the buffer format).

There could be a transition period during which wlr_output_attach_render uses this new helper behind-the-scenes to reduce breakage.

wlroots has migrated to gitlab.freedesktop.org. This issue has been moved to:

https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3079