Hello,

I am using a media player (totem) that uses the dbus GnomeSession
Inhibit call to disable screensaver.

I wrote a GnomeSession implementation that calls xscreensaver-command
-deactivate every time the Inhibit rpc is called and prints the time for
debugging purpose.

I set the blank interval to 10 min and dpms to 20 min. Screen is balnked
although the Inhibit rpc is called about once every minute. the media
player sometimes waits a bit more than one minute but definitely under
two minutes.

How I am supposed to disable the screensaver when the -deactivate
command is ineffective?

Attaching the GnomeSession implementation.

Thanks

Michal

If what you are trying to say is, "xscreensaver-command -deactivate doesn't work", you're wrong, and this is easy to demonstrate by just running it in a loop every couple minutes. Something else is going wrong with your script that is not the fault of xscreensaver.

If what you are trying to say is, "xscreensaver-command -deactivate doesn't work", you're wrong, and this is easy to demonstrate by just running it in a loop every couple minutes. Something else is going wrong with your script that is not the fault of xscreensaver.

And what that would be?

I have the log of xscreensaver-command -deactivate being called within
two minutes before the screen blanked.

I unblank it and it's right there.

Thanks

Michal

And what that would be?

I have the log of xscreensaver-command -deactivate being called within
two minutes before the screen blanked.

I unblank it and it's right there.

Thanks

Michal

Well, it does not.

It deactivates the screensaver but does not prevent the DPMS blank so
it's pretty much useless as running the script demostrates.

Thanks

Michal

Well, it does not.

It deactivates the screensaver but does not prevent the DPMS blank so
it's pretty much useless as running the script demostrates.

Thanks

Michal

It may very well be issue with the xscreensaver packaging in debian.
It says that X11 screensaver extension is present but not compiled
into xscreensaver.

Extending the script to call not only xscreensaver but also the X11
Xss extension fixes the problem.

I am not sure xscreensaver would try to use Xss in this case but since
it is not complied in it's out of question anyway.

Either way, xscreensaver-command -deactivate does not prevent screen
from blanking so it does not perform as advertised and is undeniably
broken even if you are saying it is not.

Thanks

Michal

It may very well be issue with the xscreensaver packaging in debian.
It says that X11 screensaver extension is present but not compiled
into xscreensaver.

Extending the script to call not only xscreensaver but also the X11
Xss extension fixes the problem.

I am not sure xscreensaver would try to use Xss in this case but since
it is not complied in it's out of question anyway.

Either way, xscreensaver-command -deactivate does not prevent screen
from blanking so it does not perform as advertised and is undeniably
broken even if you are saying it is not.

Thanks

Michal

No, xscreensaver should never be compiled with support for the MIT-SCREEN-SAVER extension. It's crap and breaks everything.

But now that I understand that you're talking not about xscreensaver activation but about DPMS only, I know what the problem is, and it is, as far as I know, impossible to fix portably. See the #if 0 block around line 348 of driver/timers.c:

  /* This will (hopefully, supposedly) tell the server to re-set its
     DPMS timer.  Without this, the -deactivate clientmessage would
     prevent xscreensaver from blanking, but would not prevent the
     monitor from powering down. */
#if 0
  /* #### With some servers, this causes the screen to flicker every
     time a key is pressed!  Ok, I surrender.  I give up on ever
     having DPMS work properly.
   */


So, X sucks and you are doomed.

Workaround: disable DPMS globally. Just let the screen saver paint the screen black instead of powering off the monitor. If you're using an LCD and not a CRT the difference in power consumption will probably be negligible anyway.

The difference in light output between a black screen and powered off
screen is not negligible. The backlight just leaks somehow.

It is true that using Xss requires a fork to get a sane semantics but
since xscreensaver-command is a short-lived process that should not be
a problem if the extension is used. And since the daemon is contacted
over X protocol anyway there is no reason it has to be done in the
daemon itself. And if XResetScreenSaver works that's fine too. Mplayer
seems to use it and it does not cause screen flicker. It did not work
with my test script, however. Since it returns a value I would expect
that the call is finished by the time it returns but it might need a
XFlush.

Thanks

Michal

No, that's not how xscreensaver-command works.

I'm glad that your driver does not flicker. But others do. So like I said, this can't be solved portably.

If you want to make it work on your machine and only your machine, flip the #if 0 and rebuild. But that cannot be done in general because that makes things terrible for people who have different hardware than you happen to.

Is it resetting screensaver or frobbing dpms what causes the flicker?

And what drivers do cause it?

Are they even in Debian?

Thanks

Michal

I don't know.

I don't remember.

I don't care.

It is a thing that happens, and therefore it a constraint of the world with which I have to deal. This line of questioning will not make me change my mind. I'm not putting in special-case hacks just for the system you happen to be using this week. If that's what you want, compile your own private copy and enjoy. *I* have the goal of building software that works on *all* systems, not just yours.