idle-inhibit: concerns about mandatory surface visibility
During the implementation of this protocol in labwc a concern was raised that respecting inhibition only from visible surfaces is troublesome for cases like screen sharing. For example:
- A user shares the screen with a web browser
- Opens a PDF document in a fullscreen PDF viewer
- Starts talking
The only application that knows about the need to inhibit the screen locker is hidden and the screen will be locked to the user's frustration. Been there, very frustrating. It is possible to work around with idle inhibitors built in e.g. panels and compositors, but IMO it makes the protocol less useful.
I suggest to remove the clause about surface visibility from the protocol specification.