Jonas Ådahl

The only reasonable thing I assume would to ensure that any communicated pid is the pid of the application itself, not any dbus proxy, remapped to the host pid namespace,...

I agree pids are problematic, and I don't know the intended use case, but I imagine if it was the pid pointing to the dbus proxy, it wouldn't be very...

> I'm aware of #711, so far my best idea is to have a specific method `SetKeyboardShortcut(shortcut_id)`, with the `shortcut_id` being the identifier to bind the two portals together. I'm...

There would be two levels of breaking the capture in the qemu/virt-manager case wouldn't there? As in a application level and compositor level. A compositor level escape hatch would always...

Draft backend implementation in xdg-desktop-portal-gnome: https://gitlab.gnome.org/GNOME/xdg-desktop-portal-gnome/-/merge_requests/33 It lacks libei integration, but can create barriers, get zones etc.

> Following IRC discussions with Wim, I'm changing this PR to add a new method to the Devices portal instead. The reasoning behind this shift in direction is that it...

Hmm. I see what you mean, the client would take over "ask for access" action from the camera portal/pulseaudio, by doing it "manually" via AccessDevice. Yea, maybe this is the...

Maybe we should add properties for whether a certain device class exists on the system, as currently handled by the camera portal. Quoting the Camera pull request: > In order...

> Should we change these property names? They seem fine to me already.

> How will monitor nodes be handled in this proposal? IMO the capability to output to an audio device and to capture output from an audio device should be separate....