[16:21] <alkisg> Hi, in previous Ubuntu versions, there was a patch for fusa that was hiding the reboot and shutdown menu items in LTSP clients. I think that patch was dropped in Karmic, along with the switch to indicator-applet.
[16:21] <alkisg> Would it be possible for me to suggest a smaller patch for indicator-applet, i.e. to set a xatom with the action that the user selected (reboot, shutdown etc) so that the LTSP display manager would read it and act appropriately?
[16:26] <mac_v_> tedg: ^ ;)
[16:27] <tedg> alkisg: No, but for a good reason :)
[16:27] <tedg> alkisg: We're now checking with the ConsoleKit settings to see if the user can do that.
[16:27] <tedg> alkisg: So you should be patching/fixing consolekit so that it happens all over the desktop.
[16:28] <tedg> alkisg: Including the session menu.
[16:28] <alkisg> tedg: the user dbus session is on the server, so that won't work for the ltsp client
[16:28] <alkisg> Ah, consolekit, not gnome-session... hmm...
[16:28] <tedg> alkisg: ConsoleKit is on the system dbus.
[16:28] <tedg> alkisg: And if it's on teh server, it should definitely not be giving users permission to shut it down! :)
[16:29] <tedg> Hiding the menu item is one thing, but there shouldn't be a way for them to be able to craft a dbus command to do it either.
[16:29] <alkisg> LTSP currently doesn't install policykit1 on the chroot... so if we installed it, and called the appropriate method, would that work?
[16:29] <alkisg> (I tried to call the CK method for restart, but it complained about PK)
[16:30] <tedg> alkisg: I'm not sure about PK here.  We're just calling the CK method "CanRestart" and "CanStop".
[16:30] <tedg> alkisg: I'm not sure how you configure/modify those in ConsoleKit.
[16:31] <alkisg> tedg: Anyway I got it, thanks a lot, I'll try to see how that would work with LTSP
[16:31] <tedg> np
[16:31] <tedg> mac_v_: Thanks for the ping!
[16:31] <alkisg> (i.e. I'll try to somehow make CK.restart work... thanks again)
[16:32] <mac_v_> hehe , actually i wanted to ping you for something else and caught the chat ;)
[16:32] <mac_v_> tedg: any news about inkscape ppa ?
[16:33] <tedg> mac_v_: I haven't had time to work on it :(
[16:33] <mac_v_> hmm , :( ... 
[16:33] <tedg> mac_v_: I need to migrate the branches from SVN to BZR
[16:33] <tedg> mac_v_: Well, they were already in BZR, but now that we've moved the version numbers don't line up.
[16:34] <tedg> mac_v_: I'm planning to work on it tonight, if it goes smoothly it'll be done -- but this stuff rarely goes smoothly :)
[16:34] <mac_v_> tedg: the "save" is killing me... thats the only problem ... np... take you time :)
[16:34] <mac_v_> yay
[17:10] <alkisg> tedg: could you please direct me to the package/source file that calls CK for canrestart/restart etc?
[17:11]  * tedg looks
[17:13] <tedg> alkisg: Hmm, I can't find it... I may have lied. I swear I wrote that code...
[17:13] <alkisg> tedg: heh, np, I'll look myself, thanks. Is that indicator-session?
[17:14] <tedg> alkisg: Yes.  bzr branch lp:indicator-session
[17:14] <alkisg> Thanks a lot! :)
[17:16] <alkisg> src/gtk-dialog/logout-dialog.c
[17:52] <alkisg> tedg: if I understand the sources correctly, ck-pk-helper.c asks CK/PK if the user can shutdown the system, but the actual action is in gtk-logout-helper.c, e.g. res = dbus_g_proxy_call_with_timeout (sm_proxy, "RequestShutdown", INT_MAX, &error,...
[17:52] <alkisg> The action happens on the session dbus, which is on the server.
[17:52] <alkisg> I think that LTSP needs some help on the action part, i.e. if LTSP_CLIENT is set in the environment, set an xprop for LDM to know if it's supposed to reboot/shutdown etc. Would you consider accepting such a (small) patch?
[19:06] <alkisg> As it turns out, there's already code in LTSP that reads a file named /tmp/ldm-logout-action, and if it contains "shutdown" or "reboot", it acts appropriately. So a small patch in indicator-session that would write those values to that file if the LTSP_CLIENT env var is present, would be most welcomed by LTSP users (instead of the xprop I mentioned before).
[19:10] <alkisg> (and btw, we teachers also miss the real user name instead of the login name :))
[19:14] <beuno> alkisg, I think that letting people configure their display name would be a good thing to do
[19:15] <beuno> alkisg, would you like to bring that up on the ayatana mailing list?
[19:15] <alkisg> beuno: sure, thanks.
[19:15] <alkisg> I've filed a launchpad bug for this already
[19:15] <alkisg> (some days ago)
[19:16] <beuno> alkisg, great, will help us track it. I think the mailing list is a better forum for discussions, and decisions  :)
[19:16] <alkisg> Sure, will do ;)
[19:16] <beuno> and we still have time to change it in Lucid!
[19:17] <alkisg> Lucid (and all LTS releases) are very important for schools, as they don't usually upgrade every 6 months... :)
[19:22] <beuno> yeah, there is a lot riding on Ubuntu this cycle
[20:02] <mac_v> beuno: is it really essential to show the user name in a single user setup ? :(  
[20:02] <beuno> mac_v, I don't think it's the name that is essential, but the actions under it
[20:02] <beuno> what else would you display?
[20:03] <mac_v> though it makes sense in alkisg's scenario  , its kinda wasting space on the panel ... i already know my name;)
[20:03] <alkisg> mac_v: an option for it would suffice, it doesn't need to be default... but even on my standalone PC I don't like seeing "alkisg", I prefer seeing my Greek name.
[20:03] <mac_v> beuno: IMO , just the icon would suffice , not sure why the name is needed
[20:04] <beuno> mac_v, so, you now have something to reply on the list  ;)
[20:04] <beuno> I don't think I feel strongly about it, but it seems to feel right
[20:04] <beuno> and I know there are some plans for the future in that area, that integrate/group more features under there
[20:05] <alkisg> Also, is the top panel space important? It's usually empty :)
[20:06] <mac_v> alkisg: i use only 1 panel :)
[20:07] <mac_v> alkisg: if you ask the dekstop team , they are planning on removing the bottom panel and a lot of widgets to reduce the boot time
[20:07] <mac_v> desktop*
[20:07] <beuno> I think 1 panel is the future
[20:07] <alkisg> mac_v: well, then you are not using the default setup anyway....  Ah. Ouch. :)
[20:07] <beuno> I'd have to check of it's Lucid future, or Lucid+1
[20:07] <mac_v> alkisg: they are going to make it the default or atleast try to convince UX , , since their alloted time is 4secs ;)
[20:07] <beuno> but it's near future
[20:07] <beuno> djsiegel1, do you know?
[20:08] <tgpraveen1> mac_v: 1 panel would be a pretty radical change.i doubt UX will approve.
[20:09] <mac_v> sure they wont ;p  , but lot of the applets are gonna be removed 
[20:10] <djsiegel1> I do not think we are changing anything about the panels
[20:11] <mac_v> kenvandine knows more whats going/planned to be removed ;)
[20:14] <kenvandine> beuno, it isn't set in stone or anything, but i think the goal was to try it 
[20:15] <kenvandine> and see how it impacted startup time and see what UX thought
[20:15] <kenvandine> beuno, you should really talk to rickspencer3
[20:15] <beuno> thanks
[21:17] <alkisg> tedg: adding those 4 lines to gtk-logout-helper.c make LTSP work fine with reboot/shutdown: http://paste.ubuntu.com/333393/
[21:17] <alkisg> Unfortunately, LTSP won't have client<=>server dbus integration for some time to come, so I think that that hack is the best that LTSP users can have right now...
[21:17] <alkisg> Tested, works fine, an additional check "if LTSP_CLIENT env var is defined" would be better though.
[21:18] <tedg> alkisg: Where is the gnome-session running?
[21:19] <alkisg> On the server
[21:19] <tedg> alkisg: Where is indicator-session-service running?
[21:19] <alkisg> On the server
[21:19] <tedg> alkisg: Then why is client <=> server dbus integration an issue?
[21:20] <alkisg> Because we want shutdown to shutdown the client, not the server
[21:20] <tedg> alkisg: So it seems to me the session manager should be doing that, eh?
[21:21] <tedg> I'm not trying to blow you off here, it just seems like changing the menu items is just dealing with the surface of the problem.
[21:21] <alkisg> tedg: The LTSP devs are planning to see if they can make the session dbus talk with the system dbus *on the client* instead of the one on the server
[21:21] <alkisg> But that won't happen for a long time (if it's ever going to happen)
[21:22] <alkisg> tedg: so, ltsp users now have reboot/shutdown menu items that just logoff
[21:22] <tedg> Yes.  But right now we have  "indicator-session-service -> session-manager -> console-kit" it seems like it should go to "indicator-session-service -> session-manager -> ltsp-tool"
[21:22] <alkisg> With those 4 lines you can make them actually reboot/shutdown...
[21:23] <alkisg> tedg: I understand that it's a hack. But LTSP can't do better right now.
[21:23] <tedg> ? Why can't gnome-session be fixed?
[21:24] <alkisg> gnome-session runs on the server, it has no control over the client
[21:25] <tedg> So does indicator-session-service, right?
[21:26] <alkisg> Right. So it would just send a "signal" to the ltsp display manager, so that when the session is over, LDM shuts off the PC
[21:26] <tedg> So it seems like gnome-session could send that same signal to LDM.
[21:26] <alkisg> I don't think patching gnome would be any easier...
[21:27] <tedg> Yes, but when other people use that gnome-session shutdown/restart interface it would work.
[21:27] <tedg> Like when you press the soft power button on the outside of the case.
[21:28] <alkisg> You're right on that. But I don't think I'll ever be able to get such a patch accepted.
[21:28] <alkisg> So Lucid, like Karmic, will go out with LTSP users unable to reboot/shutdown... :(
[21:29] <tedg> alkisg: Well, you only have to get the patch into the package.
[21:29] <alkisg> Well, I'll try. Thanks a lot for your time, and your advice! :)
[21:30] <tedg> alkisg: Yeah, okay, I checked and that whole interface is a distro patch anyway.
[21:30] <tedg> alkisg: So you just need to edit 95_dbus_request_shutdown.patch
[21:30] <tedg> alkisg: I believe that the guy who can help you the most there is chrisccoulson
[21:30] <alkisg> tedg: nice, you gave me hope!!! I'll try that tomorrow, too late here to go on :)
[21:31] <tedg> alkisg: He wrote that patch.
[21:31] <tedg> alkisg: Good night!
[21:31] <alkisg> Good night, and thanks again :)
[21:31] <chrisccoulson> hello :)
[21:31] <alkisg> Ah, hi chrisccoulson!
[21:31] <alkisg> Do I have any chances? :)
[21:31] <chrisccoulson> hi alkisg
[21:32] <chrisccoulson> i haven't read the scrollback yet - it might be quicker to ask your question again ;)
[21:32] <alkisg> (a small patch for LTSP users to be able to reboot/shutdown the clients)
[21:32] <alkisg> g_spawn_command_line_async("ltsp-localapps \"/bin/sh -c 'echo shutdown > /tmp/ldm-logout-action'\"", &error);
[21:32] <tedg> Instead of calling ConsoleKit in gnome-session.
[21:32] <alkisg> That command should be executed if LTSP_CLIENT is defined in the local environment, so that the LTSP display manager would then actually poweroff the pc
[21:32] <tedg> (CK would shutdown the server instead of the client)
[21:33] <chrisccoulson> this is a patch to gnome-session right?
[21:33] <alkisg> Right
[21:33] <tedg> It's a patch to your patch :)
[21:34] <alkisg> I haven't looked at the sources yet, I was trying to put it to indicator-session, but tedg corrected me by telling it's best to go to gnome-session...
[21:34] <chrisccoulson> oh, ok :)
[21:34] <alkisg> (and it indeed is)
[21:34] <chrisccoulson> yes, it's probably best somewhere like that, but i don't really know enough about how LTSP works to have much of an opinion
[21:35] <alkisg> chrisccoulson: I think I could have stgraber and ogra to verify that it works
[21:35] <alkisg> (ubuntu devs)
[21:35] <chrisccoulson> yeah, no problem:)
[21:36] <alkisg> Thanks! I'll try it tomorrow, and come back with a full "case" :)
[21:37] <alkisg> Thanks a lot to both of you, good night.