[00:50] Hey guys [00:54] Hey. === hads_ is now known as hads [01:46] Looking for tips. I VPN to a remote buntu server, and want to mount some of it's directories to the local workstation. Do I really need Samba/NFS/sshfs to do this? If so, which would your recommend? [01:48] sgrover: to access a remote filesystem as part of the local directory tree, you need a network filesystem. [01:48] sgrover: that's unavoidable. [01:49] sgrover: as you are connecting over a VPN, I presume you do not need to encrypt traffic over the network filesystem, so NFS and CIFS are valid choices. [01:50] sgrover: in general, you should use NFS for unix-to-unix networking, and CIFS if Windows is involved. sshfs is only for emergencies, as it is not an efficient nor featureful filesystem. [01:50] twb: Thanks. Just needed a sanity check before I began setting up Samba. (I need to share/connect to Windows resources from that server anyways...) [01:50] sgrover: if you already have Kerberos authentication set up, then AFS, NFSv4 or CIFS might be good choices. [01:51] Note that in all cases, you *will* need a centralized authentication protocol (e.g. LDAP or NIS) -- or you will need to manually synchronize UIDs and GIDs between hosts. [01:52] NIS is usually discouraged because -- though simple -- it is old, inflexible and has numerous security problems. [01:52] I was going to keep it simple and avoid kerberos (for now). For the most part, the server needs to mount a single windows share so a custom app can access files. Me mounting the server's directories is for development purposes.... [01:52] LDAP servers are very difficult to set up if you aren't used to LDAP. The client side is now pretty easy. [01:53] OK, if you just need to mount CIFS (Windows share), then you can put that in /etc/fstab. [01:53] I've gotten familiar with LDAP - our custom app is PHP based and uses the Active Directory server for authentication and management. [01:53] Righto. [01:54] Thanks though. Samba it is. Perhaps with AD authentication... we'll see.. :) [03:11] What's an appropriate choice to serve up filespace from an Ubuntu Server to Xubuntu workstations? [03:13] NFS ? [03:14] JanC: that could work, conceivably. [03:15] ball: on a trusted network, NFS between Unix hosts is a good rule of thumb. [03:16] twb: that was my first thought, but I didn't know if there was some modern replacement [03:16] ball: there isn't really. [03:16] Okay, well that answers that one :-) [03:16] ball: all network filesystems suck badly, but the ways in which NFS sucks aren't so important for Unix on a LAN [03:17] ball: Is it on a secure LAN or does it touch the internet? [03:18] ScottK: there's a gateway to the Internet (NAS, with almost everything blocked inbound) [03:18] Because once you get out on the big bad Internet, I'm a fan of SFTP. [03:20] ScottK: that's not a filesystem, though sshfs is. [03:20] However mounting something like /home over sshfs is not too fantastic. [03:20] Yes. [03:20] Most any form of remotely mounted home has some serious downsides. [03:21] AFS is the obvious, though hairy, choice for /home over the internet. [03:21] See e.g. hcoop.net [03:21] okay, thanks. [03:22] I've no idea how it is in Xubuntu but in Konqueror or Dolphin the fact that it's remote of SFTP is mostly hidden. [03:22] considering the upload speed of most internet connections, no network filesystem is useful for /home [03:22] ScottK: that's not at the filesystem level. [03:22] ScottK: so for example you can't cd into the SFTP directory and run "make" or "gcc" [03:23] JanC: it depends on the usage for /home, too. [03:23] JanC: I mean, a GNOME desktop uses /home very differently to a shell account on a VPS [03:24] twb: on Ubuntu (and I suppose Xubuntu) you can use gvfs-fuse and make can work fine... [03:25] JanC: that's basically the same approach as sshfs -- but that's no longer just SFTP. [03:25] twb: True. [03:25] I'm literally failing to keep my eyes open. [03:25] JanC: that's all I was saying by "SFTP isn't a filesystem" [03:25] Goodnight everyone. [03:25] twb: of course, it's jus tused as a filesystem ;) [03:25] twb: I agree, but for a lot of use cases it's close enough. [03:25] SFTP has locking issues -- not that NFS and AFS don't. [03:26] YMMV, etc. [03:26] It is relatively simple to get up and running too. [03:26] ScottK: yes, definitely. [03:26] ScottK: especially if you already use ssh everywhere [03:27] ScottK: that, and it's easy to set up "ad hoc" mounts as necessary, without local root privileges. [03:27] For my own use I have yet to hit a need for anything beyond ssh + sftp. [03:28] JanC: have you compared gvfs-fuse to sshfs? [03:28] Now I don't know if it's because I'm not using the Gnome desktop, but just some apps, but in KDE apps I can file, save as, sftp://.... and save a file from any app. I can't do that from Gnome apps at all. [03:29] I'd be interested to read a technical compare-and-contrast. [03:29] Is that typical? [03:29] ScottK: you're supposed to be able to do that in a full GNOME environment [03:29] ScottK: you might need to turn on some GNOME daemons or something [03:29] * ScottK has no idea. [03:29] ScottK: is gvfs-backends installed, and is the gvfs daemon process active? [03:31] No to both. [03:31] So that's probably it. [03:31] ScottK: that's probably why [03:32] The kid's computer has it. Thanks to aptitude why I can see a depends that needs fixing. [04:20] hi guys I need some help on fixing a problem with nagios3 I already posted this on the forums and it seems im not getting any help but here is the post on the forum http://ubuntuforums.org/showthread.php?t=1093018 [05:36] New bug: #341526 in mountmedia "Add support for Dell Driver Injection Disk" [Undecided,New] https://launchpad.net/bugs/341526 [05:53] anyone know how to get the server to work with a wifi card during install [05:55] ? [05:57] have you seen this forum post? http://ubuntuforums.org/showthread.php?t=376283 [05:57] not sure if it will help [05:59] ill take a look [06:03] ScottK: https://edge.launchpad.net/bugs/341237 is the qemu sync that awaits motu-release approval [06:04] Launchpad bug 341237 in qemu "Please sync qemu 0.10.0-1 (universe) from Debian unstable (main)." [Medium,In progress] [06:45] anybody help me with the issue im having with nagios3 [06:49] anybody help me with the issue im having with nagios3 [07:01] Ooh, qemu 0.10. [07:37] ttx: morning. [07:37] hey [07:38] coffeedude: anything for me ? [07:38] Hehehe. didn't expect you in so early local time. [07:38] didn't expect you so late local time. [07:38] Yeah. Have the patch and testing one more local rebuild. [07:38] I don't think I'll be going to bed tonight. [07:39] I'll send you the working patch while I' [07:39] while I'm still testing. [07:39] sure. [07:39] Also if you have the blurb about new features, I could use it today. [07:40] I'm working on getting that for you as well. [07:40] Great. [07:40] There is an openssl-gpl potential link in -lsass that I need to investigate too [07:41] If you need a license exception from me to link with openssl, I can manage that. I know the openssl vs. gnutls is always an issue. Hadn't thought about it affecting you before now though. [07:43] coffeedude: yep -lsass and -rpc end up linked with openssl [07:43] you're gonna make me do something about that aren't you. :-P [07:45] coffeedude: yes, I'll probably need something, I admit :P [07:46] Can I add a linking exception clause for openssl to the likewise-open license files? [07:47] ttx: work in progress patch files in route to you now. [07:48] coffeedude: that would mean rereleasing the source tarball, right [07:48] ./sigh...Maybe not. [07:49] Let me think. [07:50] need to walk around some.....been sitting too long. be back in a few minutes. [07:50] coffeedude: I'm fine with a new source tarball, but that might be complicated to release on your side. [07:50] k [08:02] ttx: Found a few minor issues with the patches I sent you. one more round I think. Shouldn't take long. [08:02] k === jwstolk1 is now known as jwstolk [09:55] New bug: #334781 in bacula (universe) "bacula 2.4.4 configuration fails" [Medium,In progress] https://launchpad.net/bugs/334781 [09:58] Sanity check: 255.255.255.252 is /30, right? [09:58] http://krow.net/dict/subnet.html [09:58] yes [09:59] Thanks. [09:59] I hate the bloody decimal dotted quad notation [09:59] if ip is .1 [09:59] er... ignore last comment :) [10:00] ivoks: actually I was going to use /31 with .2 and .3, but then I wondered if that would mean that .2 and .3 were actually the equivalent of .0 and .255 (i.e. broadcast IPs). [10:00] twb: they aren't [10:00] This is a "network" of two machines, see. [10:00] sudo apt-get install ipcalc [10:01] ehrm. well. sipcalc 192.168.0.1/31 [10:01] The cheat sheet [10:02] twb: Interesting. I like that notation a lot. [10:02] soren: I would prefer dotted hex. [10:03] C0:A8:00:00/24 [10:03] i would prefer ipv6 :) [10:03] ivoks: well sure [10:04] everything else is just workaround [10:04] ivoks: I disagree [10:04] You can use that notation for other purposes than subnetting. [10:05] well, i was refering only to subneting [10:06] Example: A company I used to work for hosted mail servers for other companies. They each had their own mail server. We dedicated a /24 subnet to each customer. The mail server was always XX.YY.ZZ.25 (IIRC). That enabled us to have a firewall rule allowing access to and from port 25 on 192.168.0.25/255.255.0.255. [10:07] ...instead of one firewall rule per customer per service. [10:08] There's obviously no way to do that with CIDR notation. [10:19] In /etc/syslog.conf, does *.* match *everything*, or only things that have a dot? [10:19] I mean, I see "auth" in the default syslog.conf and it made me uncertain. [10:21] The manpage seems to reassure me. === cjwatson_ is now known as cjwatson [11:28] since root account is disabled, where does system mail goto? === cjwatson_ is now known as cjwatson [12:27] Hi all! [12:28] <_ruben> mrwes: check /etc/aliases [12:28] <_ruben> g'day Jeeves_ [12:28] * Jeeves_ needs a hint [12:28] I edited /etc/sysctl.conf to disable ipv6 RA. However, it seems the settings are not applied when rebooting. [12:28] Jeeves_, hey mate [12:29] Anyone here to hit me with the cluebat? [12:29] Kamping_Kaiser: Hi! [12:29] * Kamping_Kaiser smacks Jeeves_ with a cluebat (sorry , cant help with the problem though) [12:29] :) [12:29] Jeeves_: block ipv6 ra with iptables? [12:30] henkjan: yuk [12:31] I'm looking for a solution, not a workaround :) [12:32] Jeeves_: if you add ipv6 yo kernel it will be. i think the problem ist that it load ipv6 module after you have tried to set ipv6 settings in sysctl.conf and ipv6 options arent availabe then [12:33] Jeeves_: offcourse you need first check that sysctl options really work [12:33] simplexio: Hmm. That could be something [12:34] sysctl works, that's tested [12:34] Jeeves_: in my system it compain that cant change nfs options but after boot process is ready and all nfs stuff is loaded it run sysctl without complains [12:35] simplexio: ./rcS.d/S15module-init-tools [12:35] simplexio: ./rcS.d/S17procps [12:35] isnt there somekind rc.local tms file where you can put your own stuff that is run last in boot process ? .. add there sysctl -p [12:35] So the options should be there [12:35] simplexio: That's too late [12:36] what ipv6 ra does ? [12:36] When booting is completed, the box allready sent a sollicitation [12:36] so it allready has it's RA-address [12:36] well offcourse, quick hack and remove module [12:37] 13:31 < Jeeves_> I'm looking for a solution, not a workaround :) [12:38] what does that ra mean in ipv6 [12:38] _ruben: ahh..ok, root: administrator is already there -- thanks [12:38] http://www.cyberciti.biz/tips/linux-how-to-disable-the-ipv6-protocol.html [12:38] that tells just howto disable whole ipv6 [12:39] http://zhadum.org.uk/2007/03/04/taming-the-edgy-eft/ [12:39] simplexio: Router Advertisement [12:41] Jeeves_: 100% way is just compile new kernel and remove ipv6 ra option from there.. im oretty sure that it there [12:42] simplexio: Don't be silly [12:42] I've got it now. Just add ipv6 to /etc/modules [12:42] Jeeves_: #ipv6 on freenode's a good place to ask about ipv6 related stuff [12:42] that way it gets loaded sooner [12:42] people there generally tend to be quite clueful [12:42] and sysctl works fine :) [12:44] Jeeves_: please document it on the wiki ;) [12:44] i usually just compile stuff that i need into kernel, and stuff that idont need i just add in modules [12:44] henkjan: Which wiki? :) [12:44] btw, what are you trying to do, accept_ra=0? [12:44] Jeeves_: wiki.bit.nl :) [12:45] Jeeves_: you can ask Jeeves! [12:45] but i like to run latest and greatest kernels. so for me compiling new kernel isnt that big problem [12:47] stockkernels++ [12:47] simplexio: When you administer 100+ servers, you'll soon start to love stock kernels :) [12:49] :) could be, or i learn howto create on apt-mirrors and own kernel packages [12:49] later is probably harder way :) [12:50] We have our own package repository [12:50] (and a mirror, nl.archive.ubuntu.com ) [12:50] but if something brakes i can only blame me [12:50] Jeeves_: Do you want to disable the automatic address generation? [12:50] simplexio: I simply blame henkjan [12:50] different scales have different needs [12:51] maswan: Yes, but I allready managed to do so [12:51] Deeps: Indeed :) [12:51] ah, ok [12:51] from our interfaces: [12:51] pre-up modprobe ipv6 [12:51] up /sbin/sysctl -q -w net.ipv6.conf.eth0.autoconf=0 [12:51] and there is no right solution to problem, only bad and even more bad [12:52] there has been what 3 working solutions to problem allready :) have like linux [12:53] this one has been working since dapper, IIRC. So we stick to it. :) [12:53] add ipv6 to modules, load ipv6 first in interface before interface is up, change kernel [12:55] I'm not sure adding ipv6 to modules covers it always. [12:55] We did that for a while, and there were some kind of issue, I forget what though.. [12:55] and yes, the inet6 interface has to come before the inet interface [12:56] in jaunty, ipv6 isn't module [12:56] wayhey [12:56] ivoks: Ah, is the autoconf disabling fixed then? [12:57] maswan: i don't know; i just noticed that it's built in, not a module [12:57] ivoks, around ? its orudie i'm at work [12:57] ivoks: Hm. Maybe we should try it out on some server then [12:58] petia: yes [12:58] am i interrupting, ? [12:58] no [12:59] ivoks: umm, so ipv6 can't be disabled in Jaunty? [13:00] after we set up the mail server i tested it the email client and used mail.mydomain.com as an incoming server, did it still use pop3 protocol ? [13:00] cemc: i'm sure it can, appending ipv6=off or something to kernel line [13:01] petia: it's impossible to tell what you have used, just based on hostname [13:01] hostname doesn't define protocol [13:02] is there a way to tell though ? [13:02] one server (servers have hostnames) support lots of protocols; http, pop3, ssh, imap, smtp... [13:02] petia: you should hire someone to set up mail server for you, since obviously you don't understand basic network concepts [13:05] petia: hostname doesn't define protocol [13:05] it's just a name [13:05] it could've been mysuperserver.domain.com [13:06] :) [13:07] right, but maybe you could have a look and let me know which hostname i should use in the incoming server according to the way the server is set up [13:07] cemc: Why would you want to disable ipv6? [13:09] <^law^> guys , i just installed ubuntu server in virtual box after that i can't boot it . it shows need pae enabled, but i can't enable the pae in my virtual box it can't be checked it is disabled. what shoud i do then ? XD [13:10] petia: server has only one hostname [13:11] petia: it can have aliases [13:11] petia: and, i repeat, hostname of the server doesn't decide what protocol your client will use [13:11] ivoks, ok [13:11] petia: in most of the cases, user that uses that client decides which protocol will be used [13:12] by selecting the protocol during configuratiob === mdeslaur_ is now known as mdeslaur [13:12] ivoks, well i know for sure that pop3 protocol is set up on the server, i just didnt try it yet [13:12] s/b$n/ [13:12] then try it [13:13] ivoks, i'm fucken scared lol [13:13] ivoks, cuase if this shit is not gonna work my boss is gonna kill me [13:13] ivoks, i'm about to point DNS to the new server [13:13] petia: please try keep the language clean ;) [13:14] petia: what did you do till now? [13:14] what did you configure? [13:14] postfix/dovecot and roundcube webmail client === `6og is now known as Kamping_Kaiser [13:15] <^law^> guys , i just installed ubuntu server in virtual box after that i can't boot it . it shows need pae enabled, but i can't enable the pae in my virtual box it can't be checked it is disabled. what shoud i do then ? XD [13:15] Jeeves_: because I don't use it (yet) and I don't want to change all my running servers (apache, ssh, whatnot) not to listen on ipv6 too ? [13:15] tested emails with outlook and thunderbird, howver i used mail.example.com as an incoming server [13:15] cemc: If you don't use it, what's there to disable? : [13:15] petia, imap is set as well, so therefore i'm not sure which protocol mail.* used [13:16] you really don't get it? [13:16] server serves protocols [13:16] i understand that hostname does not define protocol [13:16] client *decides* which protocol will be used [13:16] uhu [13:17] lol [13:17] i think i got it [13:17] i dont work on the client side too much dude [13:17] well, you should know what your clients are using, if you want to setup server [13:18] servers exists so that they could serve clients [13:18] that's their whole purpose [13:18] well if we take thunderbird for instance which is what i have on my desk [13:18] ok [13:18] in the add new email account [13:19] it doesnt ask you if your email acount is using pop3 or imap [13:19] i'm sure it does [13:19] ok i'm in the add account wizard [13:20] Jeeves_: have you looked at 'netstat -nlp' with ipv6 enabled? or 'ip a' with o lot of interfaces/IPs ? and generally I like to not run stuff I don't need :) especially on servers which are low on memory ;) [13:20] it asking, select the type of account you would like to set up : the choices i have is Email, Rss, Gmail, or newsgroup account [13:20] i go to email [13:20] petia: http://www2.cit.cornell.edu/computer/email/thunderbird/account-both.html [13:20] cemc: ipv6 does not require much more memory :) [13:20] 3rd screen [13:20] ok now i see [13:20] yes [13:21] yeah [13:21] ok [13:21] :) [13:26] Jeeves_: still I believe that what you don't need it's not supposed to run ;) [13:34] cemc: You should enter the 90's :) [13:39] Jeeves_: :) [13:41] because obviously you won't ever need to communicate with those computers which get ipv6 instead of ipv4 in a few years ;) [13:43] mark: I did not say it needs to be removed. But if I want to, I should be able to disable it ;) especially if I won't be using it soon [13:43] being able to disable is good [13:44] and I definitely will know if and when I need it, so I won't forget I disabled it :) [13:45] cemc: Just start using it! Saves you the frouble of disabling it :) [13:45] ;) === Guest98153 is now known as tfb === tfb is now known as nob === nob is now known as nobz === nobz is now known as nob_z [13:51] hi, who can help me with spamassassin sa-learn? [13:58] nob_z: Whats the problem? [13:59] spamassassin seems to not use the learned mails [14:01] I followed the install of spamassassin+postfix-dovecot+amavisd-new in the ubuntu server guide [14:02] nob_z: You needs at least 200 messages before bayes gets used [14:02] 200 of both types [14:02] ivoks, you there? the email mesages are not coming in [14:02] Jeeves_, I use more than 200, around 1000, but nothing changed [14:03] seems that I dont used sa-learn in the right form [14:03] nob_z: There is a way that you can see if bayes is being used [14:03] nob_z: sa-learn said 'xx messages read (xx used)' or something like that? [14:04] Jeeves_, yes... and sa-learn --dump magic increades the number of nham nspam tokens [14:06] nob_z: Than it's used [14:07] nob_z: And if you run [14:07] spamassassin --lint -D1 [14:07] ? [14:07] New bug: #326235 in samba (main) "Winbind crashes with new samba-package (3.5)" [Low,Incomplete] https://launchpad.net/bugs/326235 [14:07] Do you see bayes-notices there [14:07] ? [14:07] trying... [14:11] Jeeves_, yes, there are bayes-notices [14:12] Jeeves_, sa-learn must be used as root or by every user? [14:13] nob_z: That depends how you configured bayes [14:13] where I can check it? [14:17] /etc/spamassassin/local.cf ? [14:17] How can I make squid show an html page I made as homepage? [14:20] fevel: http://www.visolve.com/squid/squid24s1/miscellaneous.php [14:25] petia: yes? [14:25] ivoks, [14:26] ivoks, how can i change the output of this postconf | grep mydestination ? it needs to be a new domain, the old domain is still there [14:27] postconf -e "mydestination=blablabla" [14:31] ivoks, there was a file where i remember i specified the mydestination [14:32] petia: postconf -e will edit it [14:32] so... [14:32] postconf -e "mydestination=blablabla" [14:37] ivoks, paul@atlantalin:~$ postconf | grep mydestination [14:37] mydestination = jpiamerica.com, localhost.jpiamerica.com, localhost [14:37] now i have this [14:37] which is the correct domain [14:37] ok [14:38] looks good [14:38] is anything else needs to get changed when changing the domain address ? [14:38] i mean [14:38] when changing the domain name [14:38] hm... maybe myorigin [14:39] i think it working now :) [14:45] thanks Jeeves_ [14:46] always_direct [14:47] :) [14:50] mathiaz: okay, i tested the hardy kvm-84 packages in my ppa a bit more thoroughly [14:51] mathiaz: i've published an update to ~ubuntu-virt's ppa [14:51] kirkland: great. I'll give it a try then. [14:51] kirkland: have you made some progress on the kernel part? [14:52] mathiaz: yes [14:52] mathiaz: you can now install and run the kvm-source to get the kernel module [14:52] mathiaz: it's installable now [14:52] mathiaz: will require dkms, though, don't know if that's a problem for you [14:52] kirkland: awesome - well I'm running hardy - would it be? [14:53] mathiaz: shouldn't be [14:53] kirkland: *a problem* [14:53] mathiaz: i just reinstalled my test box with hardy amd64 [14:53] kirkland: awesome - I'll give it a try once alpha6 is released [14:53] mathiaz: running kvm-84 kernel and userspace [14:53] mathiaz: i left it running last night, testing nested virtualization [14:53] mathiaz: a virtual machine running a virtual machine [14:54] (something that was just introduced in kvm-82) [14:54] kirkland: did it freak out? [14:54] mathiaz: i was going to blog about that :-) [14:54] kirkland: are you able to run a bundle of guest inside a kvm guest? [14:54] kirkland: or is it just one-one [14:55] mathiaz: it did not freak out [14:55] mathiaz: it's been running over night [14:55] mathiaz: with the guest's guest doing debuild qemu in a loop :-) [14:56] mathiaz: no errors yet, on either host [14:56] mathiaz: it's not one-to-one [14:56] mathiaz: i think i have solution for that dovecot bug... [14:56] mathiaz: basically, the first guest can modprobe kvm, and get a /dev/kvm [14:56] mathiaz: the second guest isn't nearly as fast as the first [14:56] mathiaz: but it's not as slow as emulation either [14:56] kirkland: oh it does qemu debuild - this is why it doesn't freak out. The guest *knows* about it and is just very gentle and peaceful... [14:57] mathiaz: :-D [14:57] mathiaz: i should try mysql instead [14:57] kirkland: that will bring down your mythtv setup - be careful [14:57] kirkland: ! [14:57] ivoks: yes? [14:58] mathiaz: i added comment to the bug [14:58] ivoks: ok - I'll get to it then [14:58] mathiaz: one note about nested virtualization .... [14:58] mathiaz: it only works on amd cpus [14:58] ivoks: during my daily bug folder processing [14:58] mathiaz: sure... no rush [14:58] ivoks: thanks for following up! [14:58] kirkland: awesome - that's what I have :) [14:58] mathiaz: well... here's c/p: I suggest we check if dovecot.conf (minus ^protocols.* line) is different than the default one (minus ^protocols.* line) and, if it is, in preinst we copy that file into dovecot-postfix.conf and let the ucf do the rest. What do you think? [14:59] ivoks: hm - so why not always do that and get rid of dovecot-postfix.conf? [15:00] ivoks: even for the normal case? [15:00] i suggested that in first place [15:00] but then we would have to steal dovecot.conf from dovecot-common [15:00] cjwatson said that wasn't good idea [15:01] attempting to share or move configuration files among packages is generally a recipe for creating bugs, especially if the movement is in any way conditional, yes [15:01] cjwatson: even between binary packages? [15:02] can I set apache to listen on the same port as my proxy? they are both on the same server [15:02] configuration files are only ever owned by binary packages [15:02] it doesn't matter if they're in the same source package [15:02] you *can* move a configuration file from one binary package to another, but it absolutely has to be unconditional, and it ought to be permanent [15:03] and that's not the case here [15:03] it's also a lot of fiddly code to get right; naive approaches will fail in messy ways [15:03] * cjwatson has been there and done that [15:03] cjwatson: right. I think we're talking about modifing a config file from a another postinst binary package [15:03] mathiaz: absolutely forbidden [15:03] right, i've myself was trying to find a sane logic for ucf stealing and i didn't come up with a good one [15:04] cjwatson: even if both binary packages are from the source package? [15:04] mathiaz: (from the policy point of view, as well as sanity) [15:04] mathiaz: yes. [15:04] that makes no real difference [15:05] cjwatson: hm - well then I know a couple of package that do that [15:05] mathiaz: if they're from the same source package, then the binary package that currently owns the file can and should take steps to expose some proper interface by which the other binary package can do what it needs to do [15:05]