IRC Logs for #crux-devel Tuesday, 2008-09-30

*** Rotwang has joined #crux-devel01:32
*** DarkNekros has joined #crux-devel02:01
*** DarkNekros has quit IRC02:56
*** nipuL has quit IRC03:44
*** j^2_ has quit IRC03:45
*** Rotwang has quit IRC03:45
*** pedja has quit IRC03:45
*** pitillo has quit IRC03:45
*** Rotwang has joined #crux-devel03:47
*** j^2_ has joined #crux-devel03:47
*** pedja has joined #crux-devel03:47
*** pitillo has joined #crux-devel03:47
*** cptn has quit IRC03:50
*** cptn has joined #crux-devel03:50
*** nipuL has joined #crux-devel03:53
*** tilman has quit IRC04:04
*** tilman has joined #crux-devel04:08
*** sets mode: +o tilman04:08
*** tilman has quit IRC04:15
*** nipuL has quit IRC04:16
*** tilman has joined #crux-devel04:16
*** sets mode: +o tilman04:16
*** treach has joined #crux-devel04:18
*** cptn has quit IRC04:22
*** cptn has joined #crux-devel04:23
*** rehabdoll has quit IRC04:45
*** rehabdoll has joined #crux-devel04:45
*** nipuL has joined #crux-devel04:51
*** nipuL has quit IRC05:09
*** nipuL has joined #crux-devel05:22
*** DarkNekros has joined #crux-devel05:57
*** Rotwang has quit IRC07:29
*** DarkNekros has quit IRC08:54
*** DarkNekros has joined #crux-devel09:22
*** thrice` has joined #crux-devel11:20
*** Rotwang has joined #crux-devel11:41
*** gicoviello has joined #crux-devel11:42
gicovielloI was asking myself if we (I'm one of the cruxppc) can distribute firefox package with official branding11:44
cptndo you11:46
cptnerr, do you modify the package in any way?11:46
cptnbookmarks, default settings etc?11:46
gicoviellowe just compile your port11:46
gicoviello(crux port)11:46
Rotwanggicoviello: you distribute binary packages?11:47
gicovielloRotwang: the cruxppc iso11:47
treachdoes anyone know how far this thing goes, really?11:47
treachI mean, we're shipping a binary on the iso as well, but afaik, it's built a bit different than the offical package..11:48
cptn"You may distribute unchanged official binaries downloaded from to anyone in any way subject to governing law, without receiving any further permission from Mozilla. However, you must not remove or change any part of the official binary, including Mozilla trademarks."11:48
treachyes, yes. but what's "unchanged"?11:49
tilmanwe're not distributing an official binary11:49
thrice`also, crux compiles, not official binary11:49
treachexactly. so.. we shouldn't call it firefox then?11:49
gicovielloso we have to omit --enable-official-branding when we're compiling packages for the iso?11:49
cptnah, indeed11:49
tilmangicoviello: that would be the safe way :)11:50
thrice`I'd guess if crux gets permission to use --enable-official-branding, any crux ARCH port should be ok :)11:50
gicoviellotilman: and also the simpler ehehe11:50
tilmanbut, *did* crux get permission?11:50
treachnot afaik.11:50
thrice`I'd guess jaeger would have been careful about that, no?11:51
cptnwell, we're safe for the _port_11:51
cptnsince that's the user compiling it, not us building it11:51
gicoviellothe port is ok11:51
treachI did ask this a long time ago, but either I worded it poorly, or nobody cared then.11:51
cptnbut it's different for the iso indeed11:51
tilmanhooray, yet another reason to kill mozilla/firefox11:51
* treach votes for "earth squirrel" instead. :>11:51
gicovielloand what about asking for the permission?11:52
treachburocracy, IMO.11:52
treachnot that I have any say in it. ;)11:52
cptnwell, it would be interesting to hear what they say11:52
thrice`actually, it would be a Per question about getting permission, not jaeger,eh?11:53
treachthey changed that after per left iirc11:53
tilmanany random person can ask them11:53
gicovielloa friend of mine said to call it "fairfocs"11:54
treachok, I have no say, but IMO it's easier to just sidestep the whole mess and A) drop ff from the iso, or B) call it something different11:54
tilmanA works for me ;)11:55
tilmanwe can still distribute a xulrunner package, right?11:55
gicovielloand replace it with?11:55
tilmanfirefox compiles in <5 minutes anyway11:55
treachwell, it's not like it stays current very long anyway, right? =)11:55
thrice`I do admit, though, the nice thing about the iso is you can have a working xorg, WM, terminal, and web browser by default11:56
gicoviellothe web browser is needed on the iso11:56
gicoviello(links eheheh)11:56
treachwell, we could ship something else.. like midori/arora or what ever11:56
cptnwell, let's just disable the branding then11:56
cptnI mean, users can always go ahead and enable it11:56
thrice`or, perhaps permission was already granted, and it's not an issue? :)11:57
gicovielloi agree with cptn11:57
thrice`oh, sorry, I thought this was #crux; I'll keep to myself :)11:57
cptnalso, it doesn't cost anything to ask11:58
gicovielloagain i agree with cptn11:58
cptnunless it requires some registered entity11:58
gicoviellowho will ask?12:00
cptndo you have an idea where to ask?12:01
cptnor anyone else for that matter?12:01
cptnis there a form somewhere?12:01
gicovielloi've no idea12:01
cptnsounds like a starting point12:02
thrice`option 3) could be to distribute the binary on the iso12:02
treachthat would be B) "_or_ B)*" Unless you're suggesting we keep flying beneath the radar.12:04
gicovielloa mail to "partners at mozilla dot com" should be enough12:04
treach...I don't like the idea to have to ask permission for doing stuff with purportedly free software.12:05
* treach stands over in the debian corner, and wonders how the hell he ended up there.12:06
cptnwell, we can just replace the following in mozconfig:12:06
cptnac_add_options --enable-official-branding12:06
cptnchange this to disable12:06
cptnand we're done12:06
cptnlet's check what changes12:06
thrice`might not be bad for the iso, and then change it back in the ports?12:11
thrice`or, not at all :)12:12
cptnI don't know what the changes are when you disable it12:12
cptnis it that bad?12:13
*** gicoviell has joined #crux-devel12:13
gicoviellsorry my connection went down12:13
treachlooks like someone should volonteer to build it to see what happens. No point in 10 people doing so12:13
*** gicoviello has quit IRC12:23
*** jue has joined #crux-devel12:57
cptnhey jue12:58
*** sip has joined #crux-devel12:59
cptnhi sip13:00
cptnlong time no chat :-)13:00
sipyeah was busy mail-bombing our subscribers13:00
sipI realized I purged my apologize email from the postfix queue13:01
gicoviellyou can mail-bomb also mozilla guys then13:01
cptnso: :-)13:01
treacharen't we considered bad guys by half the internet by now? :p13:02
sipat least yahoo hates us ;-)13:02
cptnso, most of the items are already done13:03
tilmanthere's still no python or binutils release13:03
cptnwe've no branch yet for opt IIRC13:03
cptnor do we?13:03
cptnunless there's a volunteer, I can have a look at ISO building13:04
cptnit's been a while since I had a look, but jaeger promised to explain if needed :-)13:04
tilmani'll branch off opt.git13:04
cptnboth the python and binutils change probably need some testing13:05
cptnthere's also the git ports driver13:05
cptnunfortunately, I've not had the time to create an updated version13:05
tilmani wasn't sure whether the git driver really was a good idea tbh ;)13:07
tilmanbut i didn't follow the discussion closely eitehr13:07
cptnokay, let's not make that a critical item for 2.5 then13:07
sipI think rsync can handle our next scm switch13:07
sipyes I was pointing at you hehe13:08
tilmanspeaking of it13:08
tilman*cough* update *cough*13:08
* sip runs13:09
tilmanhow do we go about updating it at all?13:09
sipok I think it's better to plan a specific date for the update13:09
tilmanof course13:10
sipas I mentioned I think it's possible to live-upgrade the server13:10
sipthe only thing we need is some backup in case things go wrong13:11
sip*not as in data backup13:11
*** jkr has joined #crux-devel13:12
gicoviellthings usually go wrong13:12
jkrAhoy, important people :)13:12
tilmanwhat kind of backup then?13:12
sipsome assistance from the kalmar guys for example13:12
cptnsip: did you have any contact with Charlie recently?13:12
jkrShould I describe the bug again or does everyone here read #crux too? :)13:13
gicoviell(sorry for the intromission) but why we (ppc and x86) don't work together?13:14
treachgicoviell: -> #crux13:14
treachit's not something for the meeting, I can tell you in the other channel.13:15
tilmansip: mmh, that might be difficult to get :D13:15
tilmanbut we should try to sort that out really soon now13:16
jkrWell, whatever. I upgraded util-linux-ng yesterday and got a strange problem when booting this morning. It says that /etc/mtab doesn't exist. When the system is up, I see that /mtab is there, but my / partition seems to be mounted 8 times (at least according to mtab)13:16
cptnI'm about to commit the --extract-only thing to pkgutils13:16
tilmancptn: excellent!13:16
jkrWith version 2.13 it works fine.13:16
cptnnot that it's critical to 2.5 :-)13:16
cptnjkr: can you file a bug?13:16
jkrWhere? CRUX or util-linux-ng?13:17
jkrI've also looked at the changes in mount.t a bit and there are actually some changes in how it creates and updates mtab.13:17
jkrAh, ok13:17
jkrI could try to make a patch myself, but I don't know how to test it without rebooting every time ;)13:18
gicovielli think that you can test removing mtab and then mounting something13:19
tilmanmmh, i wonder how pointful it is to roll test iso that still have the old binutils and old pyfon13:19
jkrHmm, that could work13:19
juewrt pkgutils,  any thoughts to #336?13:20
juetilman: no, we should ship the test iso with the rc-version13:20
jue+ IMO :-)13:21
jkrgicoviell: The error doesn't seem to occur when I test it this way. When I look at the new mtab, it contains the new mount and the / mount13:22
jkrMaybe it has also something to do with the '/bin/rm -f /etc/mtab*' in /etc/rc13:23
*** DarkNekros has quit IRC13:23
tilmanjue: *nod*13:23
gicovielljkr: i've tested a bit, but i haven't found a solution13:23
*** Rotwang has quit IRC13:24
tilmansip: eventually we should probably migrate to cgit on, btw13:25
tilmansip: to escape the gitweb patching hell :D13:25
cptnI'd like cgit too13:25
tilmanat some point i tried to update our gitweb and nearly lost my sanity over it ;)13:26
sipI saw your upgrade attempt, stopped after 3 lines13:26
siplooks god, will give it a spin13:27
tilmanokay, there's opt.git/refs/heads/2.5 now13:28
tilmanand i made glib use the system's libpcre13:28
siplast semi-offtopic about server upgrade: will do it next tue at 20:0013:29
sipso you're around to help/blame13:30
tilmanfantastic :)13:30
tilmansip: thanks in advance :D13:30
sipnp, long overdue ;-)13:30
juesip: thanks13:31
cptnjkr: thanks13:31
tilmanmaybe #270 should be moved to the 2.5 deps13:32
jkrHe, are you using git for the ports just for development or will we get a ports -u with git support with CRUX 2.5? :D13:32
tilmani'm not convinced that keeping full history in /usr/ports is cool13:32
jkrWell, could create a release branch where only released ports are merged in13:33
cptngit still clones a repository13:33
cptnnot the latest state13:33
tilmani gotta run clean my shoes13:33
cptntilman: :-)13:33
jkrIsn't there an option for for only getting one branch?13:34
juejkr: why do you want that?13:34
jkrI don't want it, but I somehow like the idea13:35
jkrIt's so geeky :)13:35
sipcoolness factor: +113:35
cptnjkr: check here:
cptnjkr: there's upsides and downsides13:36
cptnright now, no one has any solid numbers if it can be tweaked to avoid keeping history around13:36
tilmanjue: do you want to commit python 6.2rcX to opt/2.5, or do you prefer to only do that once the final release is available?13:38
juethe first it think, will commit it tomorrow13:39
cptnoh, and if anyone feels like it, please feel free to build an ISO13:40
sipI started one under virtualbox but it tends to trash my disk :-/13:40
tilmani have to admit that i'm completely clueless about iso building these days13:40
juecptn: hehe, thought you want do that?13:41
juebut I can try it the next days13:41
cptnI will, but I just read that I have to patch my kernel first13:41
cptnand I won't do that during the meeting13:42
sipspeaking of the iso, any plan on adding package updates during setup?13:42
juewrt kernel-config: do we have a policy which drivers we include at all?13:42
sipjue, as in license?13:44
jueno, I mean which disk/network etc. drivers13:44
sipoh, I see13:45
sipI won't mind building a fatter kernel with more modules selected than the current one13:45
sipsometimes when I'm in a hurry I use the cdrom kernel and /lib13:46
juesip: do you feel like creating a .config for
juebased on our current one with sensible additions13:48
sipor 2.6.27?13:48
treach27 seems to be the choice of quite a few big names13:48
tilman27 eats intel nics13:49
treachonly one kind. ;)13:49
sipheh I think it's fixed13:49
tilmanmaybe. is my source there13:50
tilmandidn't check lkml :D13:50
treachtilman: I don't think fedora/suse and I think ubuntu will ship a kernel eventually that eats intel nics. :P13:50
tilmankernel = non-issue imo13:51
cptnso, FTR: firefox is called "Minefield" without the branding13:51
tilmanpeople will build their own kernel right after installation anyway ;D13:51
cptnkina silly13:51
juetilman: if we are talking about a test iso, it is13:51
tilmanyeah well :)13:51
tilmani'm not argueing against 2.6.27 if it really fixes that e1000 bug13:52
gicoviell(there is some code made by me from the 2.6.27)13:52
juewe are now at rc8, I guess we can expect a stable release within some weeks13:53
sip+1 for .27. Cue up no carrier jokes13:54
tilmanon 09-27 they said that it's not fixed yet13:54
tilman"The Ubuntu people are some of the crazier ones (should I be surprised?),13:56
thrice`now, the big question....  does blackbox get to stay on the iso? :)13:58
juesure ;-)13:59
tilmanthrice`: are you offended by that?13:59
cptnbut could we apply the text shadow patch? ;-)13:59
thrice`tilman: um, of/c, being an openbox convert ;)13:59
sipI vote for :-)14:00
jueah, I see a new (the old) maintainer of blackbox14:00
jue-> cptn14:00
jueor sip14:00
cptnwell, given that there are probably no releases coming up, I don't mind :-)14:01
tilmansip: looks like that might offend other bb users :D14:01
treachis bb really alive?14:01
treachlatest release in 2005..?14:01
tilmandoes anyone know *why* libxslt is in packages.opt?14:02
tilmanwas it because xcb-proto required it?14:02
tilmanor is there another reason?14:02
sipI think we added it to opt for 2.214:02
tilmanwhich of these three questions were you answering? :D14:02
tilmanmmh well, sec.14:03
sipsorry, was browsing the bb cvs14:03
sipanyway, was talking about libxslt14:03
tilmanChangeLog ftw14:04
tilmanwe added it in 2.314:04
tilmanwhich was the first release to come with modular xorg14:04
tilmanso it seems likely that the xcb dependency was the only reason14:04
tilmanxcb doesn't use xslt anymore (they are using python now)14:05
tilmanso i'll kick libxslt off the iso14:05
tilmanlet's see whether anyone complains :)14:05
cptnso, is there anything stopping us from building a test ISO?14:06
tilmanyes, i need to update packages.org14:06
tilmansynaptics would probably be nice to have on the iso14:07
cptnand there's the lvm2/dmcrypt issue14:07
cptnbugs 341 and 34214:07
tilmanbut i don't think that's blocking us :p14:07
tilmannot sure what the status of that is14:07
sipbtw, regarding iso building, I feel the urge of having a kind of official build machine14:08
cptnsip: that would be very cool14:08
* treach suggests sip's desktop machine.14:08
* sip looks at treach's notebook14:09
sipstarting from a always-clean setup would be really cool14:09
sipand will allow providing up-to-date packages14:10
thrice`yeah, jaeger had a net-setup (I think) that even supported the feature of fetching up-to-date packages14:11
tilmanhow did we assemble iso/setup.dependencies?14:11
tilmanall those x packages need to have new deps (s/libxslt/python/, but i think it's ordered, so. :x14:11
juemake setup.depedencies14:11
jueand make setup.dependencies-clean or make distclean14:12
tilmanoy, it's adding trailing whitespace14:13
sipoh I missed the new python dep :-/14:13
juetilman: ?14:14
tilmanjue: my new setup.dependencies file has a trailing ' ' in each new line :)14:14
tilmanor maybe git diff is playing tricks on me14:15
juehmm, looks for example like that for me:14:15
tilmanhehe, it's only hilighting the trialing space in _new_ lines, but not on old ones14:15
jueautomake: gawk db gdbm autoconf perl automake14:15
tilmanjue: nevermind :)14:15
jkrIs the output of /etc/rc logged somewhere?14:20
tilmanheh, libxcb still needs xslt14:20
tilmanso, should we also inject python? :)14:22
tilmanit's not a runtime dependency, fwiw ;D14:23
jkrmount seems to think that mtab doesn't exist everytime it's trying to update it and calls create_mtab(), which opens /etc/mtab with a+ and appends a root fs entry every time to the existing mtab :)14:23
jkr /proc/mounts is correct, by the way, only mtab is messy14:23
tilmanwhy the hell do we ship all those weird x drivers?14:24
tilmanlike glint, cirrus, chips, cyrix?14:24
sipcirrus may be used under some vm afair14:25
tilmanmmh, true14:25
sipbut I agree the others are odd14:25
cptnyeah, qemu uses it14:26
cptnor rather emulates it14:26
treachhard-line, technically vesa,vga and maybe cirrus and - in the same vein - vmware, would be enough, no? anyway sun stuff probably makes less sense :>14:28
sipopenchrome would be nice too14:29
sipeven if it's outside the official xorg tree I think14:30
cptnthat's for via, right?14:30
tilman*chrome* is for via, yes14:30
tilmanthere's video-via, there's openchrome and there's unichrome :D14:31
tilmanat least14:31
tilmanso, should i add the mach64 and r128 drivers?14:31
sipdidn't follow the latest news about via releasing source code14:31
tilmanthat's seriously old hardware14:32
cptnwow, rage 128, it's been a while indeed14:33
cptnmy first laptop had one of these :-)14:33
treachas long as it's available through the ports, is there a problem?14:34
*** gicoviell has quit IRC14:34
tilmanall those *input* drivers need to go14:37
tilmanall we need on the iso is -evdev, -mouse and -keyboard14:37
cptnsounds good14:37
tilmanplease say "ack" or "nak" now14:37
tilmanoh, and synaptics14:37
jueack :-)14:38
treachthat's input. gotta go. :>14:38
tilmantreach: yes, i'm talking about input, not video drivers, atm14:44
thrice`yes, it is very much a hassle un-checking those during an install :D14:45
tilmanyou should all use -evdev btw14:46
tilmanit replaces both -mouse and -keyboard on supported systems14:46
thrice`does it use hal ?14:47
thrice`or am I thinking of somethign else14:47
tilmandon't think so14:48
tilmani don't really know which video drivers to kick and which ones to keep14:50
juehave to go soon, do we have a real plan now how to proceed with the test iso?14:51
tilmanupdate opt/python, then hit the "build" button?14:52
cptnsounds like a plan14:52
cptnI'll try14:53
tilmanoh, look, a squirrel!14:53
cptnI'll coordinate with tilman :-)14:54
sipgreat, thanks cptn14:54
juegreat, thanks cptn14:55
jueopen question is whether to inject python or not14:56
jkrYay, I found the bug :)14:57
sipI would include it for consistency14:57
tilmanit's weird to install xcb-proto which depends on it, and then *not* inject it14:58
tilmani'll edit the setup-helper14:58
juehmm, I think we don't need that, because it's in the dep list of xcb-proto14:58
sipI'm starting to hate xcb-proto with a passion14:59
*** DarkNekros has joined #crux-devel14:59
tilmanhow come?14:59
sipwell it doubled our iso size :-)14:59
sipnah but the trend says it will, sooner or later15:00
jue~ 14M15:01
tilmanjkr: okay, talk to upstream please :D15:02
treachtilman: it was a joke. never mind me.15:03
tilmanokay, i can do that15:03
tilmani don't have a current iso handy15:04
tilmanbut iirc we're at 240 MB or so?15:04
thrice`some things left core, though :)15:04
tilmangetting down to 200 mb would be great15:04 says 247MB15:04
tilmanlet's remove firefox and xulrunner :)15:04
tilmanooooooh btw15:05
tilmanwhat about swapping cdrtools with cdrkit?15:05
tilmandidn't we ponder doing that for 2.5?15:05
tilmanaon: ping15:05
tilmanswapping blackbox for openbox on the iso would be a size reduction of almost 200KB15:08
juetilman: only inetutils is needed in inject_packages15:08
tilmanjue: okay15:08
juetilman: but you have a wm with a lot of deps15:08
juebut we can always switch to evilwm ;-)15:09
tilmanaren't all of those deps on the iso already, though?15:09
tilmanoh well, the real answer is to slim down the xorg packages on the iso :)15:09
sipwindowlab package is 18k15:11
sipdo we want to ship another X terminal btw?15:12
sipin place of xterm15:12
tilmanonly if the package is smaller ;)15:12
jueI'd stay with xterm15:12
siphehe you're starting to get paranoid about size15:13
tilmanmostly kidding15:13
tilmando you think we should continue to ship cvs, though?15:13
sipthe client? I think so15:13
sipit's not that uncommon15:13
treach"use the ports, luke" ;)15:14
tilmanmaybe in some parallel universe15:14
* tilman hides15:14
tilmangotta run15:14
juehmm, if so we should ship svn and git as well15:14
tilmanpeople will take photos tomorrow, and i need my beauty sleep :p15:14
cptngood luck15:15
tilmanjue: ugh. think of the deps! ;)15:15
siphehe good night & cheese!15:15
tilmanthanks, i'll try to hide15:15
jueyeah, so I'd remove cvs15:15
sip*sniff* ok15:16
jueleaving now, night all15:18
cptngood night jue15:18
sipgood night15:18
*** jue has quit IRC15:18
sipnight everybody15:32
*** sip has quit IRC15:32
*** thrice` has left #crux-devel15:40
jkrI sent the patch to the util-linux-ng mailing list, but I don't know how long it will take until it's avaiable through a release15:52
jkrI attached it the bug report (which can be closed now) for anyone who wants to patch mount15:53
cptnjkr: great, thanks a lot16:01
cptnmaybe we can also send a heads up to crux@16:01
jkrYou mean the channel?16:03
jkrOr the mailing ist?16:03
cptnthe mailing list16:03
jkrDo I have to subscribe first?16:04
cptnheh, yeah16:04
cptnI can send it if you want16:04
jkrOk, but it'll take a few minutes :)16:04
treachdoesn't posting from gmane work anymore?16:04
jkrNah, I can do it16:04
cptnor I can moderate you through16:04
cptntreach: it does, but one still has to subscribe16:04
cptnanyway, I'm off as well16:07
cptngood night!!16:07
*** DarkNekros has quit IRC16:09
jkrSent it to the list16:13
nipuLsorry i missed the meeting guys, forgot to set my alarm16:59
treachtimezones ftl.17:03
jkrRavioli ftw17:25
treachanyway. it must suck to be part of a "community" where everyone (pretty much) else sleeps when you're awake.17:27
jkrWhat do you mean?17:28
treachjkr: it's hard to get any real social interaction going, and it's difficult to get into the process of getting things done.17:28
jkrAre you talking about this channel or real live? :)17:29
treachI'm talking about our newly arrived friend down under just experienced.17:29
treachabout *what*17:30
jkrWoah, I would NEVER set my alarm just to get up at night and chat with some strange geeks in IRC ;)17:31
treachmeeting done, decisions made, people gone when you arrive.17:31
treachwell, that way it's really hard to get involved too.17:31
jkrI'm sure there is a time where everyone is (kind of) awake ;)17:32
treachnope, not really17:32
nipuLits not like i get up in the middle of the night for the meeting17:33
treachmost of us are europeans, but there are people from all over the place.. canada, the us, alaska, russia, australia, new zealand, you name it17:33
nipuL5am isn't too early17:33
treachnipuL: it is to me! :D17:33
nipuLbut it still sucks being up at that time17:33
nipuLwell feel free to find me a cushy job in a cushy european country so i can attend crux meetings ata  sane hour ;)17:35
treachI think I'll stay at admiring your devotion, going up at that time volontarily. :>17:36
jkrUhm, anyone using dovecot here?17:37
jkrWith cmusieve plugin?17:43
jkrDoes it work? :D17:43
jkrDid you do anything besides loading the plugin?17:44
jkrIt doesn't seem to use my script and it also doesn't log anything interesting17:44
nipuLyou need to set up dovecot to be your LDA17:45
jkrYou mean the deliver tool they mention in the wiki?17:45
jkrI thought it would do that by default17:46
nipuLno, you need to configure your MTA to use it17:46
*** treach has quit IRC17:49
nipuLin postfix for example you need, mailbox_command = /usr/lib/dovecot/deliver17:57
nipuLthen configure the lda stuff in dovecot17:57
jkrYup, already done that, but postfix seems to have a problem with the permissions17:59
jkrLooks like deliver can't write to /var/log/dovecot.log with postfix' perms18:00
jkrWhat do you have default_privs set to?18:02
jkrOh, the wiki says I have to suid deliver18:03
nipuL-rwxr-xr-x 1 root root 790080 2008-09-02 19:55 /usr/lib/dovecot/deliver18:03
nipuLi didn't18:04
nipuLdovecot is running as root18:04
nipuLthe login daemons aren't18:04
jkrYup, same here18:05
nipuLbut dovecot and dovecot-auth are18:05
jkrHowever, it's not dovecot that has problems with the permissions, it's postfix18:05
nipuLi think i have some chuser settings in dovecot.conf though18:05
nipuLi can paste my and dovecot.conf if you like18:05
jkrYeah, that'd be great18:08
jkrYou don't have a default_privs set at all18:13
nipuLi can afford to be a bit lazy, single user system18:15
jkrDoesn't seem to be the probem anyway18:15
jkrHe, seems to work with 777 perms on /var/log/dovecot.log :)18:18
jkrThis sucks18:18
jkrAh, looks like postfix calls deliver as the user who gets the mail18:22
nipuLodd, all my logging for postfix and dovecot goes to /v/l/mail18:23
nipuL-rw-r----- 1 root root 463044 2008-09-30 23:17 /var/log/mail18:23
jkrI'll just try to make deliver write a per-user log instead of the global one18:25
jkrIts messages are for the user only anyway18:25
jkrI wonder how it can work for you18:36
jkrDoesn't deliver try to log stuff?18:37
jkrAh, now I got it :)18:40
jkrOr not18:42
jkrYeah, it works :)18:46
jkrAlright, gotta sleep18:49
jkrGood night18:49
jkrAnd thanks for the help18:49
nipuLanyone know a decent file storage service? i have a few gigs of data i want to move off my vps19:26

Generated by 2.11.0 by Marius Gedminas - find it at!