As far as I know, yes. You tell me the alternative if you’ve got it.
If all you’ve got against Flatpak is it uses more storage, then I don’t know what to tell you. I have a 1TB drive that cost $80 and my GNOME system with 106 flatpaks uses just under 7%. The original post claiming 2TB is absurd.
There is no reason that you couldn’t, for instance, bind-mount the host’s nvidia drivers into the container namespace when launching the flatpak. Would avoid having to download the driver again, and reduce runtime memory pressure since the driver code pages would be shared between everything again.
And the only possible way to have that is to burn through disk space?
As far as I know, yes. You tell me the alternative if you’ve got it.
If all you’ve got against Flatpak is it uses more storage, then I don’t know what to tell you. I have a 1TB drive that cost $80 and my GNOME system with 106 flatpaks uses just under 7%. The original post claiming 2TB is absurd.
There is no reason that you couldn’t, for instance, bind-mount the host’s nvidia drivers into the container namespace when launching the flatpak. Would avoid having to download the driver again, and reduce runtime memory pressure since the driver code pages would be shared between everything again.
I don’t have the time to make a “stop doing math” meme for Unix permissions
So don’t change the defaults?