Package: src:nvidia-graphics-drivers Severity: normal On 5.12.2018 2.56, Andreas Beckmann wrote: > On 2018-12-03 15:30, Timo Aaltonen wrote: >> On 3.12.2018 16.29, Andreas Beckmann wrote: >>> On 2018-12-03 14:04, Timo Aaltonen wrote: >>>> I've packaged egl-wayland and uploaded it, but got a reject message >>>> because the binary (libnvidia-egl-wayland1) was already in the archive. >>>> Would you mind dropping it from the nvidia packaging and use this >>>> instead? This (together with the -dev package) is needed for >>>> gnome-shell-on-wayland/Mir to work with the nvidia driver. >>>> >>>> I'd then need to add the necessary B/R/C and bump the epoch, of course. >>> >>> In general no objections. Did you test whether your packages will work >>> as drop-in replacements for the ones built from the blob? >> >> No, I don't have nvidia, but the lib is their baby so I'd expect it >> would. Besides, I don't think it's used at all right now. > > There are also > > nvidia-egl-wayland-common > nvidia-egl-wayland-icd > > and > > libnvidia-legacy-390xx-egl-wayland1 > nvidia-legacy-390xx-egl-wayland-icd > > from the 390xx legacy driver. > > What should happen to them?
Gone as well. -icd doesn't seem to install anything useful anyway, so I'm not sure about it's purpose. -common just installs the json file which the new libnvidia-egl-wayland1 will include. I'll add all the C/R/P in egl-wayland, thanks. > What should we do in stable? > nvidia-graphics-drivers/stretch will probably be upgraded to 390.xx in > the next stretch point release. > The *wayland* packages in their current form were only added to stretch > in 384.111-4~deb9u1, they haven't been there from the beginning. > If we are sure they have no users in stretch, we could just remove them > from stretch again. (The packages are only recommended, not being > depended upon.) Yeah I think they can be safely dropped without any harm. > PS: Please open a bug against src:nvidia-graphics-drivers and post all > discussion there to have this archived publically. And move discussion > there. sure, this reply should open the bug. -- t