How to disable mouse pointer / prevent mouse action

[papierkorb27]

Hello everyone, I am looking for help concerning a problem using the current Raspbian OS Release with Wayland and the new default labwc. I am trying to use this setup as a kiosk mode and the product "Divera 24/7 MonitorApp" (https://help.divera247.com/display/FAQ/Monitor+Desktop+App) for fire departments. This application opens a menu sidebar at the left side of the screen, if you mouse over it. If there is a reconnect of the HDMI signal, it seems that the position of the mouse pointer activates the sidebar and the content of the screen isnt longer full visible.

Do you have any ideas how to resolve this problem with labwc? I've already seen the discussion about "HideCursor" (#2633) but as I understood it just hide the cursors but it is still at the screen position? If this should be the solution, what would be the exact steps to setup and hide the mouse pointer from startup and after a hdmi reconnect?

I would be very happy about help :-)

[tokyo4j]

I don't know how we can fix this. How did you fix it in X11 session (probably Openbox)?

[papierkorb27]

Before using labwc it worked with unclutter (https://wiki.ubuntuusers.de/unclutter/) (X11) and later in wayland (with wayfire I think so) I used hideaway (https://gitlab.com/interception/linux/plugins/hideaway) with success. Both solutions doesn't work now with labwc.

[johanmalm]

I'm unclear on the mechanism that is causing the issue, so my suggestions may not be helpful. I'm happy to help find a solution to this, but it may require you to do some of the "detective work" as your use-case is not that easy for me to setup.

Off the top of my head I feel I should mention that we support WarpCursor which can be invoked by a different process via the wtype command. I'm not sure if that would help at all because I don't understand what's causing the problem, but might be worth a try.

This application opens a menu sidebar at the left side of the screen, if you mouse over it.

Any idea what type of surface the sidebar is? For example, are we dealing with XWayland or xdg-shell? Is it a popup, XWayland override-redirect or a layer-shell surface?

If there is a reconnect of the HDMI signal, it seems that the position of the mouse pointer activates the sidebar...

Any idea where the pointer is at this point?

If this should be the solution, what would be the exact steps to setup and hide the mouse pointer from startup and after a hdmi reconnect?

We don't support triggering actions on HDMI re-connect out-of-the-box, but it probably wouldn't be hard to do by a quick hack.

With the unclutter and hideaway solutions - at what point did you trigger them? Did you link that to the output re-connection?

If it comes to it, would you be in a position to build labwc from scratch to try to patches? If it's something that does not make sense to merge upstream, would a local patch provide an adequate solution?

@cillian64 @spl237

[johanmalm]

There are some tips here for gathering information:

• https://github.com/labwc/labwc/blob/master/CONTRIBUTING.md#debugging

[johanmalm]

Depending on what's causing the issue, the following settings may be helpful too: https://labwc.github.io/labwc-config.5.html#focus

[papierkorb27]

Thank you very much for the information, I'll experiment with it. I'am new to labwc, the correct steps to testing the parameters HideCursor and WarpCursor in the correct way should be:

1. Adding to $HOME/.config/labwc/rc.xml the content from action: Add HideCursor action #2633 (comment)

2. Adding "wtype -M alt -M logo h -m alt -m logo" to $HOME/.config/labwc/autostart

After a reboot the mouse pointer should be invisible until moving it?

[johanmalm]

Yes, looks good. Let us know the result. As I don't understand the root cause of the issue it's hard to predict if it will fix it.

[papierkorb27]

Thanks, the shortcut is working now and the mouse pointer is invisble - but as expected the problem still exists. After reconnecting the hdmi output the sidebar pops up. The position of the pointer doesn't matter - after the reconnect it is still at the old position. Maybe it is not a problem with the mouse pointer and another reason (the commands by reconnecting the hdmi communication?) The problem does not exist if you start the application, just every time after the hdmi reconnect.

In the old configuration without labwc unclutter and hideaway were just started one time during the boot process and ran in the background.

There a some people looking for a solution at the facebook group of the application but at this time the working idea was switching back to X11. But in the future more people will get this problem.

How can I get the information you need about the type of surface the sidebar? I think the application is just an special overlay of the chromium browser. The application files are freely available here: https://help.divera247.com/pages/viewpage.action?pageId=119865789
Everyone can download them, if you need credentials for testing purposes please just get in touch.

MonitorApp.Image: ELF 64-bit LSB executable, ARM aarch64, version 1 (SYSV), dynamically linked, interpreter /lib/ld-linux-aarch64.so.1, for GNU/Linux 3.7.0, BuildID[sha1]=4b12022a0a60af0d4b7920deac219fcbd1d92080, stripped

Screenshot of the sidebar if you move the mouse pointer to the left side / or reconnect hdmi:

[Consolatis]

I think when detecting new outputs we move the output cursor to 0,0 which is the top-left corner. That is what might be triggering the hover effect.

[papierkorb27]

That sounds plausible. Is there any option to disable the new output detection procedure or to change the cursor position after the detection without changing the existing code? At this kiosk setup the output is everytime exactly the same (and same hdmi port), it is just powered off and on by an motion detector.

[Consolatis]

Had a quick look at the logic and my statement was wrong. What happens is that the cursor is moved 0 pixels to the top and 0 pixels to the right (e.g. relative movement rather than absolute). So this call is not doing anything other than updating some internal state. Sorry for the noise.

[johanmalm]

I've run the application and can confirm that it is an XWayland client.

I checked it by using the following config and then just alt-tabbing to get the window type.

  <windowSwitcher show="yes" preview="yes" outlines="yes" allWorkspaces="no">
    <fields>
      <field content="type" width="20%" />
      <field content="desktop_entry_name" width="30%" />
      <field content="title" width="50%" />
    </fields>
  </windowSwitcher>

I cannot get past the login prompt, and do not want to take someone else's credentials to login.

The AppImage binary does not seem to respond to any command line options like --help.
I tried to force it to run in Wayland mode (as an xdg-shell client) with the commands below, but to no avail (i.e. it persists to launch as an XWayland client):

./DIVERA247-Monitor-2.1.0-x86_64.AppImageF --no-sandbox --enable-features=UseOzonePlatform,WaylandWindowDecorations
QT_QPA_PLATFORM=wayland ./DIVERA247-Monitor-2.1.0-x86_64.AppImageF

I tried running with WAYLAND_DEBUG=1 but do not get the usual debug output. I'm not sure what the issue is there - maybe something related to the AppImage packaging.

So what's next?

1. I feel that this really needs the upstream development team to engage. I'm happy to help brainstorm a few things, but it depends on where they are going with this. For example, do they intend to port this to Wayland at some point? It maybe as simple as bundling some Wayland support when they create the AppImage. If they need to stick with X.Org / XWayland, then it should be should be easy for a developer to just build a Wayland compositor (labwc or other) and insert some printf()s to understand what's going. Or use WAYLAND_DEBUG=1 or whatever.
2. When you get the HDMI re-connect, is there a manual way to get the panel to do whatever you need it to do (like clicking on it and then clicking somewhere else). If so, it might shed some light on what's going on.

[papierkorb27]

Ok, thanks for your work. I'll try to inform the divera support team about the problem again and this discussion thread here. I think that will take time because the developer team around the monitor application is really understaffed.

After the hdmi reconnect I can just move the mouse pointer, click on any position - the sidebar disappears and everything is fine like before :-)

[Consolatis]

Another option could be to use cage rather than labwc, it sounds like its more tailored to your specific kiosk use-case and is packaged on RasbperryPI OS.

[papierkorb27]

I'll look at that. The idea was to change as little as possible as the RaspianOS defaults and using labwc but if there is no other way..

[johanmalm]

@papierkorb27 Does it matter what the position of the pointer is when the HDMI re-connect happens? In other words, does the side-bar appear even if the pointer is on the opposite side of the screen?

[papierkorb27]

no, the position doesn't matter, the pointer can be in opposite of the screen and the sidebar appear after reconnecting.