- Package:
- console-setup
- Source:
- console-setup
- Submitter:
- Petr Baudis
- Date:
- 2014-08-07 19:00:23 UTC
- Severity:
- minor
There seems to be a problem in propagation of XKBOPTIONS setup from keyboard-configuration to Xorg. My /etc/default/keyboard is (not in sync with debconf since I have slightly complex xkb setup): However, Xorg.log shows: And sure enough, I cannot use ctrl-alt-bksp in my X session to zap the X server, a rather important functionality.
The above comment is wrong nowadays.. Did you reboot after changing /etc/default/keyboard? That file is read by /lib/udev/rules.d/64-xorg-xkb.rules on boot (or keyboard hotplug). 'udevadm trigger --subsystem-match=input --action=change' might be enough to trigger an event and re-read the configuration. Cheers, Julien
Petr Baudis <pasky@ucw.cz> (15/05/2010): You need to turn off the DontZap option, see xorg.conf(5). Mraw, KiBi.
(I hope my config file is up to date, but I think it should be.) Thanks! Reboot is not a trivial matter for me on this machine, but the udevadm incantation did the trick nicely! So I suppose updating the comment in the config file and modifying the configuration script to do this after updating the configuration should be enough to fix the bug? I do that already. However, isn't that the default again now that xkb does not set the terminate option by default anymore? If not, I think that's another bug - why would you still keep DontZap at on at this point? And the keyboard-configuration setup would be highly misleading, not mentioning this requirement during the terminate option setup.
Just updating the comment. Triggering udev from keyboard-configuration doesn't really make sense (it's a layering violation; keyboard-configuration shouldn't need to know what the users for this configuration file are, whether X is installed, etc), so I think I'd change the comment to recommend rebooting instead. The X keyboard layout can easily be changed at runtime with setxkbmap anyway, and the udevadm incantation won't change the layout at the console. Modifications of /etc/default/keyboard are probably not a routine task, so I think it's ok to have them only take effect at the next boot. Yeah, DontZap is off by default, so no need to change this. Cheers, Julien
Isn't 'udevadm trigger --subsystem-match=input --action=change' totally user-independent, though? This actually does seem to me to be the proper "layer interface", it will do the appropriate thing no matter if X is installed or not and what users there are - it will just make sure they all get the updated information. It would be nice if the information would also cater for more advanced users who would want to do without rebooting (after all, those are the most likely to edit config files directly) - but if you think that would make the comments too confusing, it is not so big deal I suppose. Thanks a lot for looking into this, Petr "Pasky" Baudis
Hi. I've basically the same problem,... running sid,.. rebooted... Config-file: BMODEL="pc105" XKBLAYOUT="de" XKBVARIANT="nodeadkeys" XKBOPTIONS="compose:caps,terminate:ctrl_alt_bksp" However, ctrl+alt+backspace doesn't kill X Any ideas? Cheers, Chris.---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.
Please file your own bug, and provide: - udevadm info --export-db - Xorg.0.log - xkbcomp :0 - Also make sure it also happens on a bare X server (no desktop environment). Cheers, Julien
Do you think it will be useful if keyboard-configuration provides a /etc/foo.d directory where the interested packages (console-setup, X) can install scripts to reconfigure the keyboard? Anton Zinoviev
It seems overkill to me, fwiw (I think it's reasonable to say that changes to /etc/default/keyboard take effect at the next boot), but if you want to provide this then I have no objection. Cheers, Julien
Personally I don't have an opinion about this. I guess I am going to wait for some more bug reports requesting 'dpkg-reconfigure keyboard-configuration' to actually reconfigure the keyboard. :) Anton Zinoviev
Hello, Anton Zinoviev, le Tue 14 Sep 2010 13:24:59 +0300, a écrit : Well, I happened to expect it. At least the attached patch should be applied since restarting udev won't fix anything, while udevadm would save people a reboot. Samuel