[Debian-eeepc-devel] Bug#598097: Make detect_x_display go away to solve this

[Ben Armstrong]

OK.  I need some input from other developers on the team ...

On 13/11/10 08:20 AM, Ben Armstrong wrote:
> After discussing this with a member of the X team, they advise this is
> simply a bad idea (to try to talk to X from acpi scripts).  I've heard
> this advice before and given the difficulty we've had, I have to wonder
> if it isn't time to finally take that seriously.  After all, this key
> sends an XF86Display event, so the WM should bind that to do whatever
> the user wants.  I think in GNOME and KDE, there should be some agent
> that already handles this.  For non-gnome/kde, something could be
> cobbled together with xrandr the way we do already, only not running as
> root.
> 
> I'm going to see what other keys (if any) rely on detect_x_display, and
> see if we can apply the same approach.

There is the touchpad toggle key.  We could use acpi_fakekey to send
XF86TouchpadToggle to /dev/input/eventx there.

I think the other key we need to deal with is the display toggle. I
don't know how to handle that. I don't see any appropriate key listed in
/usr/share/X11/XKeysymDB. Is there some non-X way to do this that will
work in X as well? I seem to recall we had some difficulty with this one
but I no longer recall the details ... Currently, this key does the
following:

handle_blank_screen() {
    if [ -S /tmp/.X11-unix/X0 ]; then
        detect_x_display

        if [ -n "$XAUTHORITY" ]; then
            xset dpms force off
        fi
    fi
}

This is ugly because it only handles the user logged in on DISPLAY :0
(just like all other uses of detect_x_display).

Ben