input device matching for forced on outputs
Regarding commits 8dc6db8c and 90718170, they would seem to contradict each other in specific situations.
I'm faced with an issue of an unreliable hotplug interrupt which wont be fixed in hardware. The current solution for displaying an image is to use the force-on ini file directive to force the display on when weston thinks that the connector is disconnected. This works well and we see an image fine while weston thinks that it is disconnected.
However the touchscreen attached to the display is not associated with the display due to 8dc6db8c. Given that 90718170 was implemented after 8dc6db8c, is it reasonable to assume that the previous restriction of touchscreen just wasn't considered when adding the force-on directive? or is this deliberate for some reason?
If it was deliberate, it would seem to be at odds with the comment in the commit log for 90718170:
Forcing connectors on is useful in special circumstances: avoid output configuration changes due to hotplug e.g. with KVM switches, or hardware with unreliable connector status readout for example.
Is handling this situation something that we would like to fix upstream? Either by allowing force-on to mean any input device associated with it is always connected, or maybe a force-on style directive for a specific input device section in the ini file?