I am writting my own wayland compositor for AMDGPU (RISC-V assembly... running on x86_64). Now, I am thinking more about it, multi-display handling MUST be don

I am writting my own wayland compositor for AMDGPU (RISC-V assembly... running on x86_64). Now, I am thinking more about it, multi-display handling MUST be done at the rendering toolkit level, certainly not with some blurry upscaling of wayland application surface. A toolkit to handle multi-display properly, must get those properties (on top of the wayland color surface depth/format): physical size, typical view distance, PPI (Pixels-Per-Inch). Dunno yet if wayland does handle all of them in order to tell the rendering toolkit. Basically, the rendering toolkit/application needs those in order to create the "right wayland surface" for an output and configure its rendering properly to this very output. If I recall properly a wayland application is notified of much of all this. An output could be a "merged" set of displays with the same critical "properties". In the end, a wayland application should, for each of its surfaces, care of only one "output". If many "outputs" with very different core properties get to "scan-out" the same wayland surface --> omega dirty mess.