<qyliss>
The code looks fairly portable to crosvm too
<MichaelRaskin>
Oh nice (re: vsock)
<MichaelRaskin>
vilhalmer: well, X11 protocol could also be implemented with each client having an illusion of being alone, unless specially authorised… if wl_roots has some plans to implement filtering, that might be interesting of course.
<MichaelRaskin>
Then we get Wayland access control done in an incompatible way between wl_roots/KDE/Gnome, right?
<MichaelRaskin>
Hm nice, so Firecracker virtio-vsock is even usable for dynamic multiplexing.
<MichaelRaskin>
So such code should be usable for a network-front VM accepting connections from dynamically created network-using VMs
<Shell>
MichaelRaskin: access control is semi-intentionally not part of Wayland, as xdg-desktop-portal is intended to be the way folks access access-controlled desktop things these days. nobody thought anyone would want to access-control the clipboard though, I imagine. putting the access control stuff in the compositor (or a subprocess thereof) for now seems fine I guess?
<qyliss>
I'd surprised nobody thought of access controlling the clipboard, since Qubes does it...
<qyliss>
And I'd have assumed Qubes would have come up as an example of access controls in desktop environments
<Shell>
qyliss: it could also have been a case of "we need a clipboard to be usable and cba doing access control yet, ship it".
<MichaelRaskin>
If anything needs access control, clipboard is in top 3
<edef>
qyliss: right, but i wanted vhost-net
<edef>
qyliss: like, purposefully
mearon has quit [Quit: WeeChat 1.7.1]
mearon has joined #spectrum
<qyliss>
edef: I don't
<qyliss>
But yes, I do understand that.
<qyliss>
crosvm's implementation is vhost-net
<qyliss>
So you get best of both worlds :)
<qyliss>
I've been finding it difficult to concentrate for the past couple of days so I think I'm going to take a few days off. Probably won't be on IRC much until then. :)
<qyliss>
(that's addressed to the channel, not a response to anybody in particular)