On Mon, 2007-08-13 at 00:31 -0400, Stefan Monnier wrote: > I believe that has no influence on my main problematic case: > 1- the user hits the power-button because (s)he wants to hibernate the > machine. I agree, unless you remove the switch, there's nothing you can do to prevent the user from pressing it a few seconds..
> 2- each and every user logged in with Gnome gets to process the request via > its own gnome-power-manager, isn't it wonderful? You're right, but i assume it was the purpose of dbus's system bus : broadcast to every listner. Maybe you can remove /usr/share/gnome/autostart/gnome-power-manager.desktop so only interested user add g-p-m to it's own session?(i never tried). > > Of course, if you don't want anybody to hibernate|suspend your computer, > > you can disable it (see /etc/hibernate/ram.conf and hibernate.conf) > > Will it really work? Is g-p-m really going to use hibernate rather than > some other method? Is that documented somewhere? As you mentioned in a previous main, it is unclear how g-p-m, hibernate uswsusp[s2ram+s2disk],acpi-tools, dbus, hal interacts. my understanding is : - g-p-m is the user front-end (including automatic reaction to events, like "battery critical" => hibernate...) - hibernate-package know what has to be done before and after hibernation AND sleep, like (the package's name is misleading) - uswsusp knows the tricks to actually susend|hibernate each and every single computer types (a machine has to be white-listed to work). - hal : "hawdware abstraction layer" as it's name implies.. but it also seems to duplicate s2ram (less `locate 20-video-quirk-pm-lenovo.fdi`) - dbus : just communication between components (hal and g-p-m) - acpisupport does a good job at converting laptop-specific function keys into standard keyboard event. However, it seems to also overlap hal's hack, hibernate-script's hack and uswsusp's hack. All i know is that a broken /etc/uswsusp.conf prevents hibernation, and a non-whitelisted computer (T61..) works if i force it in /etc/hibernate/ususpend-ram.conf (so both are needed). Franklin -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]