GNOME Bugzilla – Bug 710012
Wayland: Accessible mouse-moved events are not emitted
Last modified: 2021-07-05 10:45:44 UTC
Created attachment 257123 [details] accessible event listener Steps to reproduce: 1. Launch the attached event listener in a terminal. 2. Move the mouse pointer. Expected results: Accessible mouse-moved events would be printed in the terminal. Actual results: Accessible mouse-moved events are printed in the terminal only in a non-Wayland session. If the session is Wayland, there are no mouse-moved events.
(Sorry to be spammy, but I forgot to include a use case.) Orca has a "mouse review" feature for users with low vision. You move the mouse over something and Orca will present it. This requires Orca to be notified when the mouse moves. Without the event emission, this feature is completely broken. This feature is turned off by default btw.
Briefly looking at this, I'm once again shocked by the absurdity of the a11y stack :-(. at-spi-registryd runs XQueryPointer in a 20ms timeout whenever a "mouse listener" is registered. The relevant sources here are at-spi-core/registryd/deviceeventcontroller.c and at-spi-core/registryd/deviceeventcontroller-x11.c
If you want to keep some of the things in the deviceeventcontroller interface working under Wayland, the registryd needs a privileged interface to the compositor that lets it track mouse positions and possibly also warp it. Has somebody from the a11y side looked into this ?
(In reply to Matthias Clasen from comment #3) > If you want to keep some of the things in the deviceeventcontroller > interface working under Wayland, the registryd needs a privileged interface > to the compositor that lets it track mouse positions and possibly also warp > it. > > Has somebody from the a11y side looked into this ? Not extensively, or at least I haven't. Do you know if there is any precedent yet for creating a privileged interface / infrastructure for creating them?
(In reply to Mike Gorse from comment #4) > Not extensively, or at least I haven't. Do you know if there is any > precedent yet for creating a privileged interface / infrastructure for > creating them? Weston has its own way of doing this: launch the privileged client directly from the compositor, and pass it the wayland connection as an fd. And then only expose the privileged interface to that fd. After talking to Jonas and Ray, this is not a great fit for gnome-shell. We don't want the shell to launch clients directly like that. Instead, we should just make this a dbus interface (like we've done for e.g. monitor configuration), and not allow sandboxed clients to talk to it.
The 2 items at: https://fedoraproject.org/wiki/Wayland_features#accessibility:_mouse mention this additional protocol to be added to gnome-shell/mutter. I've filed a bug against mutter to that effect: https://gitlab.gnome.org/GNOME/mutter/issues/9
GNOME is going to shut down bugzilla.gnome.org in favor of gitlab.gnome.org. As part of that, we are mass-closing older open tickets in bugzilla.gnome.org which have not seen updates for a longer time (resources are unfortunately quite limited so not every ticket can get handled). If you can still reproduce the situation described in this ticket in a recent and supported software version, then please follow https://wiki.gnome.org/GettingInTouch/BugReportingGuidelines and create a new ticket at https://gitlab.gnome.org/GNOME/at-spi2-core/-/issues/ Thank you for your understanding and your help.