[00:00] liw: you don't speak fluent HTTP? [00:00] * ion_ is quite happy with his own server that sends the queries to the source. [00:00] cjwatson, he just tried but missed the tags :) [00:00] cjwatson, _I_ do, after implementing it for Kannel, but, er, um... [00:00] well, yeah, as it happens I do too but :) [00:01] ogra: no tags in HTTP, that's HTML :) [00:01] well :) [00:03] slangasek: well it does return the response as non-authoritative [00:03] I've had to battle with enough weirdness in DNS even when things work according to spec (my brain find DNS to be illogical), I really wouldn't want to have to deal with more broken DNS implementations than I have to... [00:03] the logical next step is for them to run a POP3 server on the same IP, and start remotely deleting your email for you [00:03] calc: that's, er, meaningless to almost all local resolvers [00:04] calc: because *every* reply that doesn't come directly from the authoritative nameserver is supposed to be tagged that way [00:26] is it normal that libtoolize --copy --force deletes config.{sub,guess} but doesn't restore them ? [00:26] not traditionally, but who knows with the new libtool :) [00:26] slangasek: I'm running into this problem when trying to build openldap on intrepid [00:26] slangasek: config.{sub,guess} are copied before autogen.sh is run, and then the build fails [00:27] that really sounds like a libtool bug to me, then [00:27] Keybuk: ^^ ? [00:28] mathiaz: did you see my reply/branch on the cn=config stuff, btw? [00:29] slangasek: yop - I've got a reply ready in my Outbox. [00:29] slangasek: I've come up with a solution to *not* bring back the perl code :) [00:29] slangasek: I've also cherrypicked some of the changes you've made in your bzr branch [00:30] wow, I'm interested to see how you managed that, then. :) [00:31] slangasek: well - just a for loop - but not that I think of it, it doesn't support includes in included files... yet [00:32] does it handle merging of continued lines? [00:32] slangasek: such as ? [00:32] mathiaz: slapd.conf supports line continuation with \ [00:32] slangasek: on includes ? [00:32] mathiaz: so if you don't handle this, you won't necessarily have parsed the file correctly [00:33] on any directive, sure... [00:33] slangasek: hm - then my solution won't work :/ [00:34] slangasek: I'll dive into that once I've uploaded something in intrepid [00:35] ok [00:41] mathiaz: try adding --install [00:51] cjwatson: thanks - works well now :) [01:15] mathiaz: pushed a couple more updates to the cnconfig branch; debconf template updates, basically [02:22] slangasek: great - I've merged your change in my branch [02:22] slangasek: I've also fixed the multi-line parsing stuff for includes [02:23] slangasek: I've pushed all of it to my lp branch [02:23] slangasek: which patch should I send for review: the one against your bzr branch or the against debian svn ? [02:58] <__keybuk> mathiaz: probably a bug? try --install [02:59] <__keybuk> ah yes [02:59] <__keybuk> don't use --force without --install ;) [02:59] <__keybuk> it won't do what you think [03:35] this new openjdk that is going to be in 8.10, is it fully open source? [03:35] because i heard openjdk still had some close elements in order for it to build [03:44] If it's in Main, it's fully open source. [03:53] or its a bug [03:56] is there a java plugin associated with openjdk? [03:57] Is another hardy point release coming in the October area? [03:57] sry if I doubled that [03:57] pwnguin: There's no such bug, although not every feature present in sun-java6 is implemented (although 97% are) [03:57] lukehasnoname: More likely December. October is for intrepid. [03:58] figures [03:58] well, bedtime for me, I'll be on in... 9 hours. [04:05] persia: january, rather [04:06] slangasek: Early January? [04:06] well, we'll see :) [04:07] Ah. I was sorta guessing 8.04.1 + six months, and (posibly mis-) remembered that being a few weeks ago. [04:07] depends on how well we think we can get the dominoes set up before everyone goes on vacation in December [04:07] 8.04.1 was beginning of July [04:08] I guess. Last year-end seemed fairly active: I'm not sure how this year will turn out. [04:17] persia: Where fairly active == two major Soyuz breakages? [04:17] wgrant: Well, there was that, but I also remember there being a bunch of people who were working on stuff: perhaps it was just extra visible because of the Soyuz breakages. === Igorot is now known as Knightlust [06:37] Good morning [06:38] ion_: so you really did have a runaway process? [06:39] or is it a bug in deluser? [06:41] hey pitti! [06:41] hey Hobbsee, how are you? [06:41] pitti: good :) [06:41] * StevenK waves to pitti [06:41] pitti: i found the pieces of paperwork i wanted, so \o/ [06:43] pitti: Neither. A bug in something that is supposed to remove a USER_PROCESS entry from /var/run/utmp. [06:44] pitti: I changed the USER_PROCESS byte in utmp to DEAD_PROCESS with a hex editor and then deluser worked fine. [06:46] Good morning pitti. [06:46] pitti saying good morning is really I sign I should go to bed. [06:46] Ditto [06:46] ScottK: heh, sleep well! [06:47] hiya dudes and dudettes [06:47] * Hobbsee stomps on Mithrandir's feet in greeting [06:48] evilHobbsee [06:48] :D [06:48] Mithrandir: That's redundant. [06:48] * Hobbsee MUHAHAHAHA [06:48] * Hobbsee tickles and pokes ScottK with the Long Pointy Stick of DOOM!!!!!!!!!!!!!!!™ [06:48] ScottK: I work for the Redundant Department of Redundant Redundancy. [06:49] Sounds like the government [06:49] njet [06:52] Fortunately ScottK has had too much Scotch to feel any of it. [06:52] Let's try a pool cue, then [06:53] Let's not. [06:53] It could be fun [06:55] I've just fixed an RC bug in Lenny, so I'd appreciate a break. [06:56] well, you may well get the break. it would probably be your kneecaps, though. [06:57] Yeah. Many StevenK will just sponsor the NMU upload instead. [06:57] Many/Maybe [06:57] Will I? :-) [06:58] * ScottK double checks to make sure he really uploaded to mentors. [06:59] Apparently, but that doesn't mean it actually shows up there. [07:04] Since mentors appears to have eaten his upload, ScottK tosses out http://kitterman.com/debian/xmms2tray_0.4-1.1.dsc in the hopes that perhaps some DD present (StevenK maybe) will pick it up. [07:04] It was an easy one to fix. [07:04] * ScottK stumbles off to bed. [07:21] doko: is debian/patches/javazic-fixup.patch still required? it's not applied anywhere [07:21] doko: I guess it's just noise (I already elminitated the sharutils b-dep and the debian/copyright amendment), but I want to confirm with you first [07:24] pitti: no, not needed anymore [07:25] doko: ok, thanks; where do these three additional javazic/tzdata_jdk/* files come from? [07:25] there's no documentation in the changelog, nor upstream (bug/code) references [07:30] pitti: part of openjdk as well. openjdk/jdk/make/sun/javazic/tzdata_jdk/* [07:31] ah, and they are not shipped in -headless, so that they need to be copied? [07:35] yes, I only did want to add the code only. this is data, which could change more often as well? [07:36] so whenever a timezone is split, the changes need to be applied there as well? [07:38] well, it's an additional symlink, isn't it? [07:39] right === tkamppeter_ is now known as tkamppeter [08:15] mvo: good morning [08:15] good morning pitti [08:15] mvo: some problems with your p-apt SRU, I mailed you [08:16] pitti: yes, I noticed. thanks. I will re-upload in some minutes [09:20] pitti: the python-apt stuff is reuploaded with the requested changes [09:20] mvo: danke sehr! [09:38] assertion: config files that include non-active lines make upgrades harder. [09:45] if nouveau didn't rely on unstable kernel interfaces and was sufficiently stable versus nv, would ubuntu ship it? [09:47] pwnguin: Most likely. As I understand things (and I'm not a nouveau expert) libdrm is currently the main blocker. [09:51] pitti: I managed to make Jockey show the 3 supported drivers instead of just one, however the result is far from perfect (and I feel that it's just a cheap workaround) : http://albertomilone.com/Screenshot-Hardware-Drivers.png [09:52] tseliot: in terms of code hten? the screenshot doesn't have anything surprising [09:52] (unless one driver is supposed to be enabled0 [09:53] pitti: the entries in the screenshot don't have the "(VERSION)" beside the name [09:53] oh, indeed [09:53] pitti: thanks for moving some of the GNOME srus to -updates (and thanks to pedro who did some testing rounds yesterday) [09:54] tseliot: I might be able to give you a hand there? [09:54] pitti: let's talk in private, I have some code to show you [10:08] persia: well, im curious about why people might feel accepting money is more dangerous than shipping code [10:09] pwnguin: I don't understand what you mean [10:27] Hello :) I've one question. did you updated the non-free nvidia driver to the current version? "http://www.nvidia.com/object/linux_display_ia32_173.14.12.html" "Fixed a problem with missing rendering in OpenGL Workstation Overlays. " that maybe fix the bug with the kde 4.1 desktop performance? [10:51] * pitti summons our nvidia driver expert tseliot [10:51] LOL [10:52] * pitti hugs tseliot [10:52] * tseliot hugs pitti [11:15] hmmm. why don't keys work anymore, on intrepid? [11:15] Hobbsee, we decided keys were too much work to maintain so we disabled them [11:16] * Hobbsee beats cody-somerville [11:16] * cody-somerville is hurting. [11:17] Agent admitted failure to sign using the key. [11:18] looks like https://bugs.edge.launchpad.net/seahorse/+bug/201786 [11:18] Launchpad bug 201786 in seahorse "ssh Agent admitted failure to sign using the key" [Unknown,Confirmed] [11:25] hmmm. maybe it isn't, as the workaround doesn't work for me [11:29] Hobbsee: http://bugzilla.gnome.org/show_bug.cgi?id=544554 [11:29] Gnome bug 544554 in general "ssh agent doesn't work correctly" [Normal,Reopened] [11:29] Hobbsee: if the issue is the ssh agent [11:31] seb [11:31] seb128: i'll live in hope... [11:32] Hobbsee: use ssh-add meanwhile? [11:34] seb128: that looks broken too. i've installed the pinentry stuff, so we'll see if that fixes it [11:34] Hobbsee: how is that broken? [11:36] couldn't connect to agent [11:39] Can you help me with customizing a livecd? I've made this sexy /etc/gconf/schemas/local-panel-default-setup.entries and to try to load it as default panel with this command: gconftool-2 --config-source=xml:readwrite:/etc/gconf/local.xml.defaults \ --direct --load \ /etc/gconf/schemas/local-panel-default-setup.entries [11:39] But the livecd starts up the same God damn standard panels every time [11:40] Why? :) === ember_ is now known as ember === sabdf1 is now known as sabdfl [12:28] pitti: I think the language-pack scripts need to be changed to take /usr/share/locale/xx/entry.desktop from kde-l10n-xx instead of kde-i18n-xx, where can I find that? [12:33] Riddell: the files are actually copied in the langpack-o-matic tree, not taken out of the packages (they are not exported) [12:33] Riddell: lp:langpack-o-matic has a directory extra-files/ with all the extra KDE bits (entry.desktop, charset, flag) [12:34] Riddell: those need to be unpacked, the .desktop replaced, and re-tar'ed [12:34] Riddell: I can do that for you, if you give me the new set of .desktop files? or want to do it yourself? (should just be a moderately complex bash for loop) [12:45] * Riddell grumps as his internet dies for the third time today [12:45] pitti: http://www.kubuntu.org/~jriddell/tmp/desktop.tar.gz [12:46] Riddell: just bear in mind that this won't actually become active until we get the first intrepid export from Rosetta [12:46] (which we all desperately wait for...) [12:46] Riddell: will apply, thanks [12:47] pitti: I was wondering what would be your take on bug 253032 -- also do you think one day it should get integrated to network-manager rather than using a separate applet ? [12:47] Launchpad bug 253032 in likewise-open "likewise-open-gui needs a better menu item name" [Low,Confirmed] https://launchpad.net/bugs/253032 [12:48] pitti: flag.png and charset seem to have disappeared, so should be just entry.desktop now [12:48] ah, that's good to know; I would have kept them [12:50] Koon: hm; knowing nothing about AD myself, "Likewise AD settings" is not much more helpful to me, I'm afraid [12:51] Koon: what does that app do? [12:51] does it just set the AD server name/workgroup, or something like that? [12:52] pitti: it allows you to join or leave an Active Directory domain (accepting new logins that are authentified in AD) [12:52] if so, that should eventually become a n-m plugin, yes [12:52] "Likewise Active Directory Membership" is a little long [12:53] I don't think we'll find a name which is both descriptive and short [12:53] the entire concept is just too complex to explain in a menu entry, IMHO [12:53] since users shouldn't even care about concepts like AD or pam-ldap, etc. [12:54] pitti: we can remove "Likewise" since likewise-open is in main as the preferred way of handling AD membership [12:54] Koon: let's do a step back [12:54] Koon: do you need that AD membership to log in in the first place, or to access network services? [12:54] (since I suspect the former, too) [12:55] pitti: the idea is to let users defined on the domain log in, so the former [12:55] Koon: shouldn't that be exposed in the login manager rather? [12:56] Koon: i. e. (just purely as a "how it should look like), gdm displays a list of available domains and allows you to pick one? [12:56] or, if it's just one, just use it? [12:56] pitti: makes sense. Though in Windows joining the domain is also done as a logged-in user [12:57] Koon: hmm; I don't understand where PAM should get your credentials from if it isn't already connected to the AD? [12:57] or is the concept based on the assumption that the user db is present locally? [12:57] (wouldn't that make the entire idea pretty worthless?) [12:57] pitti: you can both use domain logins and local logins [12:58] I understand it as the windows pendant of pam-ldap [12:58] I used the latter, but not the former [12:58] you use a local login to set up AD membership, which plugs pam modules, nsswitch etc. [12:58] from then on you can log in as a domain user [12:58] but ICBW (my windows networking knowledge is about this ---><---- big, sorry) [12:58] pitti: closer integration of AD into login would be the subject of a nice spec I think [12:58] hm, working as two different users? [12:59] that sounds overly complex to me? [12:59] Koon: what does the app actually need to do to join a domain, do you need any authentication to even connect to the server and become able to locally log in users? [12:59] pitti: that's the way it works on Windows [12:59] pitti: AD is LDAP+Kerberos so yes, it's roughly where pam-ldap is (it can interoperate with some configuration). [13:00] so setting up a network with AD requires the admin to set up local logins for every user in addition to maintaining the AD? [13:00] pitti: when you click "join" (in the elevated-privilege Likewise applet) it prompts you for a AD domain admin user/password [13:00] (well, there's probably some sort of replication going on, but it still seems weird to me) [13:01] pitti: no, no need for local users for each domain user [13:01] Koon: how do users log into a machine then? certainly not through a shared account? [13:01] The LDAP part provides a user database. [13:02] pitti: likewise hooks into nsswitch to accept users from local passwd/group and remote AD-provided ones [13:03] exactly as an external LDAP thing [13:03] (I actually meant on windows, where you mentioned the two-stage login, but ok) [13:04] on Windows you log in with the local admin defined during install, join the domain then domain users can log in and you get two admins, a local one and a domain one. [13:04] both in the "local admins" group [13:04] Koon: ah, hang on; "prompts you for a AD domain admin user/password" -> so that's *not* the user's name/password for login, but the AD admin's password for setting up this machine to allow AD user login in the future? [13:05] i. e. something that is done once as a setup, not every day as normal user login? [13:05] pitti: yes [13:05] aaah [13:05] (sorry I'm being unclear) [13:05] that's where my misunderstanding came from [13:06] so yeah, integration into gdm doesn't make too much sense then [13:06] in current hardy/intrepid I'd say that it belongs into network-admin [13:06] pitti: better as an additional tab in network-manager [13:06] but that's going away [13:06] since n-m replaces network-admin, it should go there, yes [13:07] pitti: ok so we should spec closer AD integration, but for Intrepid+1 [13:07] pitti: for Intrepid do you have any preference ? I mean, anything more descriptive than "Likewise" is OK ? [13:08] Koon: I'd aim for consistency here; ATM, the large majority of entries in the Admin menu have descriptions without names [13:08] i. e. it says "Printers" and "Hardware drivers", not system-config-printer and Jockey [13:08] so what about "Active Directory Setup"? [13:09] it's a little confusing because you aren't setting up Active Directory. You're setting up your membership to an existing AD [13:10] I think I prefer "Active Directory Membership" -- the two functions the applet can do is "Join" and "Leave" [13:10] "Active Directory Membership"? [13:10] Koon: right, "Membership" WFM [13:11] Koon: thanks for the heads-up! [13:11] persia: as in "setting up my integration inside an AD domain" [13:11] pitti: many thanks for taking the time to bear with my explanations :) [13:11] Koon: Right. It was meant as a suggestion, but you suggested the same thing at the same time :) [13:11] persia: great minds alike ? === thekorn_ is now known as thekorn [13:16] * pitti sighs at lightweight bzr checkouts; painfully slow... [13:18] hmm. is this overflow /tmp/ thing a new feature in hardy? [13:18] Ng: no, it's been there a little longer; feisty? [13:18] seb128: hmmm. running ssh-add first seems to fix the problem [13:18] the key-based auth works after that [13:19] pitti: interesting. I only just came across it, someone's laptop was full, so the deleted a load of files, but the 1MB overflow /tmp/ was full too, so things still gave out-of-space errors (specifically thunderbird, which uses /tmp/ extensively) [13:19] Ng: ah; well, it's just meant to allow you to login in the first place to clean up [13:20] there should be some kind of notifictaion, right [13:20] pitti: unfortunately the laptop in question is some thousands of miles from me, so I didn't see whether it notified or not, but evidently the user in question didn't notice it, or understand it :/ [13:21] perhaps it should trigger a similar reboot-required thing to kernel updates, so there is a more persisent notification? [13:26] it would warrant a panel icon, yes [13:26] it would integrate nicely with liw's system cleanup app [13:27] yay! more files to delete! [13:28] I added a "clean" hook into the friendly-recovery menu too - this should be integrated with the system-cleanup too probably [13:28] system-cleaner: because files are tasty and unlink(2) is faster than gzip(1) [13:34] is there a way to get diffs beetween a generic -ubuntuX and ubuntuX+1 revision? [13:34] i mean, a place where _all_ revision were keeped, or their patches (in latter case i know the answer: no) [13:35] gaspa: launchpad has those diff now, so for recent version the package summary on launchpad [13:36] seb128: ok, i'll see. [13:37] gaspa: also http://patches.ubuntu.com/by-release/atomic/ubuntu/ [13:38] cjwatson: ah, i didn't know... thanks. [14:23] seb128: gpm d-bus activated? that would be really weird... [14:24] pitti: why? it has a /usr/share/dbus-1/services/gnome-power-manager.service [14:24] oh, really! [14:25] but I thought that would be a g-p-m backend, not the applet [14:25] well, I'm confused [14:28] pitti: what do you mean? [14:28] pitti: gpm is what is used to know if you can suspend, hibernate, etc [14:28] seb128: I usually thought applets are started with autostart .desktop files, not via d-bus [14:28] yes, but I'm not speaking about the applet there [14:28] the applet is fine [14:29] that g-p-m itself which doesn't have the right environment [14:30] ah, I thought daemon and applet are the same process [14:30] * pitti doesn't find a separate power-applet process on his system [14:31] seb128: hm, so at least my session dbus does have the $XDG_SESSION_COOKIE; it might not pass it to activated services? [14:31] pitti: there is no applet, it uses the notification area and yes that's gpm itself which doesn't that [14:31] pitti: right, dbus activation doesn't pass an environment, a patch has been commited to dbus git recently to provide a way to do that [14:32] pitti: that's one of the reasons why we didn't active gnome-settings-daemon over dbus, otherwise the applications started using multimedia key can't connect to gnome-keyring, etc [14:33] *idea* [14:33] right, I remember [14:33] pitti: thanks pitti, that talk gave me an idea about the issue [14:34] seb128: so, would it help to have an autostart .desktop file for g-p-m? [14:34] gdm has an autostart directory [14:34] /etc/xdg/autostart/gnome-power-manager.desktop, hm, there is [14:34] but I put something in there to get a command line handy start on the gdm screen [14:34] and that didn't work [14:34] gdm might not look at /etc/xdg/autostart/ [14:34] ? [14:34] I guess for whatever reason it doesn't manage to autostart those [14:35] and g-p-m get started over dbus when queried [14:35] pitti: no it doesn't [14:35] hah; that would be it then [14:35] it looks to /usr/share/gdm/autostart [14:35] well, I guess that's a feature [14:35] well, there is nothing in this one [14:35] I'm wondering if that's the issue [14:36] since even though it is a session, you might not want *all* the user stuff in gdm [14:36] maybe we should have gpm there [14:36] right, I think you should [14:36] but why the upstream tarball doesn't install an autostart there then? [14:36] that's also a little less brittle than relying on d-bus activation env passing [14:36] maybe Fedora symlinks the one from /e/xdg/autostart? [14:37] doesn't seem so [14:37] at least not in the gdm spec [14:39] * seb128 grrrrs at soren [14:40] virt-manager still doesn't work in intrepid [14:44] seb128: kvm is also a bit on the crashy side for me in interpid, not sure if that is kvm or the kernel though [14:44] mvo: well, virt-manager doesn't allow to select an iso, the "next" button in the wizard does nothing [14:45] * pitti currently uses only kvm directly, which seems to work fine [14:45] meh, different problem I guess, for me kvm (directly without virt-manager) behaves unstable (crashes in the guest etc) [14:45] sometimes, not always [14:45] pitti: the fc9 beta iso I have on the disk indeed has a gnome-power-manager.desktop in the gdm autostart directory ;-) [14:45] with virt-manager, it is unbearably slow (theory: not using /dev/kvm) [14:46] seb128: Qapla'! [14:46] * pitti hugs seb128 [14:46] * seb128 hugs pitti [14:46] what is the rpm equivalent to dpkg -S file? [14:46] seb128: see, that should get filed upstream, Fedora didn't properly contribute patches back *cough* [14:46] ;-) [14:58] seb128: rpm -q something, --help usually worked for me [14:59] seb128: however, IMHO it's debatable whether this symlink shuold really go into gpm or gdm [14:59] pitti: rpm -q -f worked [14:59] I'm talking to upstream about it [14:59] argh, too many 'manager's in the session [14:59] nothing that does the real work :-P [14:59] Rename them all 'kit' and it'll be fine. [15:02] ScottK: *grin* [15:04] seb128: oh, was the -q -f successful? which package ships it in fedora? [15:05] pitti: gnome-power-manager [15:05] ok [15:12] Riddell: quite a few language desktop files got dropped, though [15:12] and 5 added [15:12] Riddell: committed and pushed [15:12] pitti: yeah, there's not as many completed langagues in kde 4 yet, I guess you can keep the entity.desktop files from KDE 3 though [15:13] entry.desktop rather [15:27] moin === mvo_ is now known as mvo____ === mvo____ is now known as mvo === mvo is now known as mvo____________ === mvo____________ is now known as mvo__ === mvo__ is now known as mvo === hunger_t is now known as hunger [15:58] hello! [15:58] is this a bug report channel??? [15:59] I don't know if is really a bug... [15:59] I run under ubuntu intrepid ibex, but I cannot usa tty1 to 6 [15:59] no problem with graphics [16:02] LocutusOfBorg: ubuntu uses a bug tracker, located at launchpad.net [16:02] LocutusOfBorg: Also, the Intrepid discussion channel is #ubuntu+1 [16:02] LocutusOfBorg: i don't think you should be running ibex... [16:03] Hobbsee, now I run under both 8.04 and 8.10 [16:03] is only for nug huntung... :) [16:03] mkrufky, I had already submitted a bug on launchpad... [16:03] https://bugs.launchpad.net/bugs/253661 [16:03] Launchpad bug 253661 in ubuntu "Intrepid Ibex bug" [Undecided,New] [16:04] ubottu, yeah [16:04] Sorry, I don't know anything about yeah [16:04] LocutusOfBorg: you should leran to file better bugs than that. put a more descriptive title on it, for a start. [16:04] oh, I know [16:04] thanks... [16:05] But my english is not good, and I didn't find a better title... :P [16:06] https://bugs.launchpad.net/bugs/253661 [16:06] try "tty 1-6 does not work with a Ati Radeon 9600 pro" or something [16:06] Launchpad bug 253661 in ubuntu "TTY! to 6 Intrepid Ibex bug" [Undecided,Incomplete] [16:06] either way, this is not a support channel, so please continue this in #ubuntu+1 [16:07] ok... [16:09] Hobbsee, title changed... :D [16:10] thanks! [16:10] have a nice day! [16:10] i'll go on the other chan [16:31] kirkland: I'm looking at ecryptfs to steal some knowledge about auth-client-config; how come that there is no postinst/prerm to trigger the PAM updating through a-c-c? [16:33] doko: ping [16:34] kirkland: the a-c-c manpage suggests that this is necessary? [16:34] well, ask ... [16:34] doko: can you make me an admin on openoffice-pkgs also? [16:35] sure [16:35] thanks :) [16:35] jdstrand: "auth-client-config -L" doesn't have 'pam-login'; does that mean that I cannot use it to update the 'login' profile? or will that Just Work once I actually add a profile for login? [16:36] jdstrand: context: for a spec of mine I need to make libck-pam-connector to add itself to /etc/pam.d/login on installation (and cleanup on removal) [16:36] jdstrand: however, I'm still pretty unsure how to use a-c-c for that [16:37] pitti: hmmm, currently auth-client-config only deals with common-* (and nsswitch.conf) [16:37] how do i delete a package that hasn't built yet from a ppa? [16:37] jdstrand: I tried to add it to common-auth, but that applies to gdm sessions as well then, which screws up [16:37] i thought i needed admin access to the group, but that doesn't appear to be enough [16:38] jdstrand: would it be difficult to make it work for login as well? if so, I can cook my own solution in the postinst, but I'd much rather use a-c-c as common infrastructure [16:38] pitti: there isn't a technical reason why auth-client-config couldn't handle arbitrary /etc/pam.d/* files [16:38] pitti: it just currenty doesn't. I'd need to extend it [16:38] eek, except that /etc/pam.d/login is a conffile [16:39] while common-session isn't [16:39] urgh [16:39] pitti: you planning to do this automatically from postinst? [16:39] jdstrand: ok, seems I can't use a-c-c for this anyway [16:40] pitti: why is it specific to login? [16:40] jdstrand: seems I should rather modify the default conffile accordingly and make it silent if the package isn't installed [16:40] slangasek: I don't want it for {g,k}dm sessions [16:40] since then you'll get two CK sessions for the same TTY, which confuses PolicyKit and others, and breaks [16:41] pitti: the only way to make it silent right now is to use the 'missingok' module option, which is a horrible kludge that I was planning to get rid of as soon as I had the PAM framework stuff implemented [16:41] jdstrand: modifying other package's conffiles is a no-no, that's why I can't [16:41] slangasek: hm; would "the PAM framework stuff" include a scenario like mine? (just apply to VT logins) [16:41] pitti: yes I know, I was trying to ascertain if you were trying to do this via packaging, or have the user run a-c-c [16:41] how does that cause two CK sessions for the same TTY, and couldn't that be addressed in either the PAM module (making the module a no-op for X sessions) or in the DM (deferring to PAM)? [16:42] pitti: no, not in the least :/ [16:42] the framework is for managing the "common" files only [16:42] the per-service files are conffiles, deliberately [16:42] pitti: based on your comments, it sounds like the former [16:42] slangasek: I guess that depends on which order the two CK sessions are created [16:43] jdstrand: just packaging, since we want to install it by default [16:43] (in which case, a-c-c is not a good fit-- unless you plan to have it own login) [16:43] jdstrand: no, that would be bad [16:43] pitti: well, couldn't your module just check pam_get_item(PAM_TTY) and only start a session if it's a real TTY? [16:43] right [16:43] (as opposed to an X display) [16:44] slangasek: if that's possible, sure; also, if gdm registers the CK session before PAM does, the pam-ck could just check for one [16:44] but if it's the other way around, it gets hairdy [16:44] hairy [16:44] slangasek: unfortunately there is no call for modifying an existing session, such as adding an X11 display [16:45] but it seems you both recommend to do use common-session and fix it in the PAM module itself [16:45] ok, I'll look into that then [16:45] jdstrand: so assume I use a-c-c with common-session, a-c-c still needs to be called in the postinst, right? ecryptfs-utils currently doesn't seem to do that [16:45] pitti: correct [16:45] I don't think you want to just check for an existing session; I'm not sure that gdm keeps enough PAM state around until the end to let the PAM module know it didn't open it in the first place [16:45] a-c-c isn't meant to be called in a maintainer script either :/ [16:46] pitti, slangasek: there is documentation in the a-c-c source packge about ideas for using it in packaging, but there is no policy in place [16:46] pitti, slangasek: though I believe there are enough options to safely do it... [16:47] slangasek: hm, so you neither like changing the default conffile to include it, nor using a-c-c in the postinst? [16:48] what's the recommended way to set things up then? [16:51] pitti: https://wiki.ubuntu.com/PAMConfigFrameworkSpec; you work on fixing your PAM module to handle the TTY, and I'll work on this :) [16:52] slangasek: ok :) [16:53] slangasek: wrt your Spec-- it doesn't handle nsswith.conf (obviously, it's a pam spec). do you have thoughts on dealing with nss too? [16:54]