That only solves one single use case, that of a single local desktop user accessing the console of a guest running under qemu:///system instance.
The goal is that desktop virt should not use qemu:///system at all in the near future, instead using qemu:///session, at which point all VMs would be running as the user's own UID in the first place.
For better serial console access in general, we want to add a API to libvirt, so that apps don't need to access the files directly at all, whether local or remote, whether privileged or unprivileged.
In parallel, we're also aiming to provide a way to tunnel the character devices over the VNC console connection.
I don't really want to add chown'ing as a short term hack for this one use case that we're aiming to deprecated, but rather concentrate on the long term viable solutions.
That only solves one single use case, that of a single local desktop user accessing the console of a guest running under qemu:///system instance.
The goal is that desktop virt should not use qemu:///system at all in the near future, instead using qemu:///session, at which point all VMs would be running as the user's own UID in the first place.
For better serial console access in general, we want to add a API to libvirt, so that apps don't need to access the files directly at all, whether local or remote, whether privileged or unprivileged.
In parallel, we're also aiming to provide a way to tunnel the character devices over the VNC console connection.
I don't really want to add chown'ing as a short term hack for this one use case that we're aiming to deprecated, but rather concentrate on the long term viable solutions.