[arch-dev-public] Mesa, Nvidia and libglvnd (the good, the bad and the ugly) :-) -second try-
For the past few releases nvidia has supported the installation of several libgl without hacks (currently nvidia-libgl and mesa-libgl). Using some patches from upstream and Fedora, we are planning to introduce libglvnd support in mesa-17.0. What does it mean? Mesa and Nvidia will be able to live together in perfect harmony. Mesa and Nvidia will no longer provide mesa-libgl and nvidia-libgl, which will be obsolete. The new libglvnd package will provide libgl support and will depend on an opengl-driver, which mesa and nvidia will provide. * The nvidia-340xx driver will still work but will probably need a specific (non libglvnd?) mesa version. * The nvidia-304xx driver will be moved to unsupported. * Bumblebee will need some patches (1). Prime is also supported as a workaround (2) When? The first step is a new xorg-server version (1.19.1-3) with upstream patches to improve outputclass support and without the glx.so symlink hack (3), now using ModulePath xorg option (4) and adapt Nvidia and Mesa drivers. Once Mesa 17.0 is released, the next step will be the libglvnd move... (1) https://github.com/amonakov/primus/issues/193 (2) https://devtalk.nvidia.com/default/topic/957814/linux/prime-and-prime-synchr... (3) https://bugs.archlinux.org/task/26284 (4) man xorg.conf -- Laurent Carlier http://www.archlinux.org
participants (1)
-
Laurent Carlier