Power button acpi action is bad for gnome (powerbtn.sh)

Agreed. Just commenting out everything so that people can use it as an example
seems reasonable.

The action of shutting down the system seems perfectly reasonable and
predictable to me.

Shutting down would make sense, if it was done properly. However, just calling
shutdown isn't a good way of closing down a system if it's being used as an X
workstation. At the very least it needs to end all gnome sessions cleanly so
that they are saved and will be restored at next boot; or else every time
someone hits their power button by accident they're going to shoot someone. You
can see how this is done for KDE in the acpi script, but I don't know of any way
to do this properly for gnome.

Sebastien, is there some way to signal GNOME to logout from an external program?

(In reply to comment #4)
> Sebastien, is there some way to signal GNOME to logout from an external program?

No real idea on this for the moment but I'll try to search for it soon.

ok, "gnome-session-save --kill" should be ok for this (thanks Vincent)

Sounds good...Thom, how about it as a TRLS feature?

Having just added myself to the group of people who have hit their power button
by accident and lost work because of it; could the priority on this be a little
higher? Since it seems I can't remember to change it every time even just on my
computers...

(In reply to comment #7)
> Sounds good...Thom, how about it as a TRLS feature?

totally missed this chunk of thread. Yes, this seems reasonable.

acpid (1.0.4-1ubuntu5) hoary; urgency=low

  * Ensure the power button kills the gnome session properly. (Ubuntu: #1403)

 -- Thom May <email address hidden> Thu, 17 Mar 2005 17:54:36 +0000

Is this update still queued somewhere? breezy is still using 1.0.4-1ubuntu4
four months on.

Apparently pbbuttonsd causes the logout dialog to appear in GNOME; should we do
the same with acpi?

(In reply to comment #12)
> Apparently pbbuttonsd causes the logout dialog to appear in GNOME; should we do
> the same with acpi?

To clarify this, pbbuttonsd does not cause the Gnome dialog; I recently changed
pbbuttonsd to not do any power button action at all by default in order to not
interfere with the Gnome dialog. Seb, Matthew, which component is actually
responsible for spawning the dialog?

Ok, this nasty behaviour has returned in Dapper. Hitting the powerbutton and
having it just shutdown is a VERY bad thing. It should bring up the logout dialog.

I cannot test this. On powerpc it does what it should, and I don't have an i386
notebook. Please, can somebody who can actually test this take this bug?

Pitti, what exactly do you mean by "does what is should"? Does that mean it
shuts down the machine when pressed or that it brings up the logout dialog?

(In reply to comment #16)
> Pitti, what exactly do you mean by "does what is should"? Does that mean it
> shuts down the machine when pressed or that it brings up the logout dialog?

Oh, sorry. It brings up the logout dialog, as requested in several other bugs.
We had the double-action bug (dialog AND immediate suspend) for pbbuttonsd too
in earlier breezy.

As said in the previous comment, I cannot test this bug since I only have a ppc
notebook (which works fine). I give this bug back to debzilla.

I agree with Corey here ! It should popup the logout dialog and nothing else.

On my computer, the button is very close to my knee and I can't count the number of time I've lost something by just watching my computer shutting itself down.... (I hear some laugh in the background, please have respect for my sadness.)

This bug is dedicated to the unknown lost data...

I'm adding a comment to this bug because I'd like clarifcation -- I noticed after upgrading to Dapper this system just powers off when I press the power button. In Breezy, pressing the power button initiated a shutdown procedure. I have no idea whether it was "good" from an X perspective, but at least it didn't just kill the power.

I noticed Bug #2882 describing behavior in /etc/acpi/powerbtn.sh but it seems pretty definite that script is NOT being invoked when I press the power button on this (eMonster 550r Desktop) system.

Is mine more likely a bug in the ACPI scripts, or an issue requiring a kernel parameter or other recognition of the power capabilities in the BIOS, or both?

This is working for me in current dapper, provided that gnome-power-manager ist installed and running. Pressing the power button invokes the logout dialog just fine.

Thanks for the reminder -- I recently updated Bug 39499 because I discovered I could make my system power off properly by adding acpi=force to the kernel parameters. AND it turns out when I do that, the power button behaves as expected, too. (This is using the 686 kernel.)

Under the 386 kernel, the power behaves as expected without adding any kernel parameters.

Hi, was this bug fixed in dapper? if so, could someone close it?
thanks.

Last time I tested it the problem still exists using the 686 kernel on my machine. Using the 386 kernel, it brings up the dialog and shuts the system down properly. Using the 686 kernel the powerbutton kills the power.

My workaround for this is to add acpi=force to the boot parameters when using the 686 kernel. Which I have to redo in grub's menu.lst after every kernel update.

I suspect it's a hardware-specific bug, but I have not tried it on other hardware to narrow it down.

This is fixed for me in dapper.