Listen to org.freedesktop.ScreenSaver events

[oz123]

This fixes #57.

[spixi]

This would also require fixes in mate-power-manager and mate-session-manager.
https://github.com/mate-desktop/mate-power-manager/blob/937ef297a9b1552c044958e69e10183aa658e5bf/src/gpm-screensaver.c#L35
https://github.com/mate-desktop/mate-session-manager/blob/fa1bf6c2fb45ff0263465537a0730ebff2b2010d/mate-session/gsm-manager.c#L95

[lukefromdc]

I don't really have a good way to test this given both my setup and workflow

[oz123]

I can build deb packages for you if it helps.
Testing is very easy. Start netflix or youtube in your browser, setup screensaver to start after 2 minutes.
You will notice that despite streaming video, the screensaver starts after 2 minutes.
When installing the packages with the patch, the screensaver will not start if you play video.

[lukefromdc]

I do not have ANY streaming accounts but when playing video in things like MPV I've never had trouble with the screensaver. HOWEVER, I do not have the available bandwidh (no landline) for long format video in browser (from a remote source) plus I normally set screensavers for very long timeouts, so this could exist but never trigger in my workflow. Building these is not the problem for me to test, not at all. Rather I have to shut down all my other work, switch from the mate-wayland session back to the default xorg session, reconfigure my screensaver settings, and then do something like playing a local video in the browser.

[oz123]

Thanks for the detailed explanation. As it seems, you are currently the only member with merge rights, so it's up to you.
I am able to set a virtual machine for you to connect with a VNC so you can test this.

[lukefromdc]

I will NOT be able to do a remote VM connection, never done anything like that. If others test this and get good results I can still merge it, thankfully. This is another example of why I cannot solo maintain MATE, we need to get more people on the team but I don't know how to do that, determine who can be trusted etc.

[mmueller]

I've had this bookmarked for a while, and I just noticed the request for testers. I tried this on my system (Manjaro 24.2.1), built mate-common and mate-screensaver in ~/mate and ran mate-screensaver manually out of there. All of my other MATE packages are running out of the OS's maintained versions.

It works, but there is a quirk.

• I set screensaver to kick on after 1 minute.
• Power management is set to turn off display after 30 minutes.
• If I idle for a minute, screensaver kicks on (fade to black then lock).
• I confirmed that with my system's mate-screensaver, it still fades to black during the video and locks.
• I killed the system mate-screensaver and ran my local build.
• Now, if I idle while watching a Youtube video, screensaver does not kick in. (Nice!)
• However, the quirk: If I pause the video and start doing anything else, about 3 seconds later, the screensaver activates. No fade out, just sudden black/locked screen.

I tested pausing with the trackpad and clicking, and using the keyboard (spacebar) and then typing away in another window, and it will literally lock the screen on me while I'm scrolling or typing.

Since I'm clicking and typing, I would expect the session to be considered active, not idle. I wonder if there's some race condition between "session is idle" flag being updated and the "screensaver is being suppressed right now" message from dbus?

It's not really the fault of this PR, but it's probably exposing something that has been wrong for a while. I'm curious, @oz123 do you see this behavior, too? Or maybe it's specific to something about my setup.

[lukefromdc]

THANK YOU for testing! We now know this needs work before I can merge it. The bug you found is a nasty one and if it got into a release would require an emergency fix and re-release

[spixi]

Please note, that we still have to listen for org.mate.ScreenSaver, because many third-party application will check for the DESKTOP_SESSION variable and will only send the org.mate.ScreenSaver events if they detect MATE.

[oz123]

Please note, that we still have to listen for org.mate.ScreenSaver, because many third-party application will check for the DESKTOP_SESSION variable and will only send the org.mate.ScreenSaver events if they detect MATE.

Can you be more specific? Which many ?

Firefox\Chrome is not anymore.

[spixi]

Please note, that we still have to listen for org.mate.ScreenSaver, because many third-party application will check for the DESKTOP_SESSION variable and will only send the org.mate.ScreenSaver events if they detect MATE.

Can you be more specific? Which many ?

Firefox\Chrome is not anymore.

You can blame me that I introduced this into PCSX2. Other projects can be found here (and this is only the list of projects which host their code on GitHub):
https://github.com/search?q=org.mate.ScreenSaver+DESKTOP_SESSION&type=code

[oz123]

Oh. Crap. All those should just list to org.freedesktop.ScreenSaver. I can give it a try to listen on multiple event types.

@mmueller Unfortunately, I can't confirm the behavior you are describing...

[spixi]

Oh. Crap. All those should just list to org.freedesktop.ScreenSaver. I can give it a try to listen on multiple event types.

@mmueller Unfortunately, I can't confirm the behavior you are describing...

Sometimes, the logic is more hidden, but yeah, we now have a chicken egg problem and should listen to both events. https://github.com/videolan/vlc/blob/5857e58efdf5aad7c20bbe19be897f2865053174/modules/misc/inhibit/dbus.c