[00:00] where can I get more info on motu-school? === superm1 is now known as superm1|away [00:18] zul: ping [01:02] Adri2000: yep === superm1|away is now known as superm1 [01:21] zul: what happened to the fail2ban sru? [02:14] Adri2000: dont know Ill have a look tomorrow [03:45] I would like to put python-django through FreezeExceptionProcess [03:46] what's a good bug title? [03:46] s/title/summary/ [03:48] would "intrepid should include python-django_1.0" work? [03:49] "FFe for 1.0" sounds about right. [03:50] RAOF: I'll use that. Thanks. [04:11] kirkland, ping [04:12] superm1: howdy! how was the 10K? [04:12] hi kirkland. well unfortunately i didnt get to run it [04:12] superm1: bummer :-/ [04:12] i broke my ankle 3 weeks before [04:12] superm1: jesus H [04:12] superm1: that sucks [04:12] kirkland, yeah during one of my training runs i slipped on a curb [04:12] twisted my ankle, there was a loud *pop* and next thing i know i'm in ER getting some xrays [04:13] one chip and one break :( [04:15] kirkland, being that you've touched sysvinit a bunch, I was hoping you might be able to help explain/figure out what's going on with this crash report: http://pastebin.com/f2cc4c86e [04:16] superm1: that sucks so much, so sorry to hear :-/ [04:16] * kirkland looks [04:16] kirkland, yeah i've come to terms with it, but i was pretty heartbroken. there's tons more races coming in oct, so i should hopefully be out of the cast in 2 more weeks and able to start training again [04:16] superm1: yeah, i hear yah.... [04:17] superm1: if it makes you feel any better.... [04:17] superm1: i trained for the 2007 Marine Corps Marathon in Washington DC last year, over the course of 6 months [04:17] superm1: 4 weeks before the race, I was in the top condition of my life, ready to run the race of my life [04:18] superm1: i did a 20 mile run on a Saturday morning [04:18] superm1: and on Sunday, I should have stayed home and watched football [04:18] superm1: instead, I went rock climbing with friends [04:18] kirkland, oh man, i see where this is headed [04:18] superm1: if that wasn't enough, I had to go jumping off of a crazy rope swing [04:18] superm1: and if that still wasn't enough, I had to climbing higher and higher into the tree to get the swing of a lifetime [04:19] superm1: well, i swung out so far, i landed in a sandbar [04:19] superm1: landed in about 3 feet of water [04:19] superm1: sprained my ankle and knee [04:19] superm1: heartbroken, of course [04:19] superm1: no break, thankfully [04:19] kirkland, ouch, how much time between then and when the race was supposed to be? [04:20] superm1: exactly 4 weeks to the day [04:20] superm1: i couldn't put weight on it for the first week [04:20] superm1: the next 2 weeks I hobbled around [04:20] superm1: week 3, I ran the IBM 10K in a lot of pain [04:20] yeah i'm just starting to put weight on mine at the 3.5 week mark but still hobbling [04:20] superm1: week 4, I "ran" the marathon, with my entire shoe taped [04:20] superm1: i was an hour slower than my goal time [04:21] superm1: and ran in a lot of pain, but I finished :-) [04:21] kirkland, but you still finished though on a sprained ankle/knee [04:21] i hope no permanent damage? [04:21] * kirkland really goes look at the pastebin now [04:21] superm1: no, i don't think so [04:21] superm1: i did a half-marathon in April and had a lot of fun, no pain [04:21] superm1: i'm just starting to think about the Austin Marathon which is in February now [04:22] superm1: i haven't decided if i'm doing it or not, yet [04:22] kirkland, that's good. i'll admit i half though of just dosing up on vikaden and trying to run through it, but realized that was a bad idea since i'm just starting to be weight bearing :) [04:22] superm1: yeah, just save it until you're 100% [04:22] superm1: you'll enjoy it so much more! [04:22] well if you are thinking for feb, let me know what training group you go with. depending on how my healing it looking, maybe i can try to shoot for a half marathon at that time with you [04:23] superm1: the Marine Corps marathon was different, because I had already bought plane tickets, hotel reservations, my mom and dad were going, i was running with a friend, etc. etc. etc. [04:23] lots of peer pressure [04:23] oh wow [04:23] superm1: that sounds good ;-) [04:23] superm1: but you're right, there are a dozen 10K's in Austin [04:23] maybe more, actually [04:24] well that and 5k's, so a nice healthy mix that i'll have a variety to look forward to [04:24] superm1: so are you thinking the problem is with # [04:24] update-rc.d: symlink: No such file or directory [04:24] # [04:24] dpkg: error processing mythtv-backend (--configure): [04:24] # [04:24] subprocess post-installation script returned error exit status 2 [04:25] superm1: or somewhere further up? [04:25] kirkland, that's what i'm thinking is causing the crash [04:25] but i don't recall ever changing any of the update-rc.d code in the init scripts for intrepid [04:25] superm1: let me pull the source to mythtv-backend [04:27] kirkland, i believe all that the postinst ends up doing (after debhelper) is update-rc.d mythtv-backend defaults 24 16 >/dev/null [04:27] kirkland, which is identical to the behavior that dh_installinit added in hardy too [04:29] dh_installinit -a -u'defaults 24 16' [04:30] yeah in debian/rules, i suppose that's what propogates the update-rc.d into the postinst. [04:30] should that be different then? [04:31] superm1: looking at one that I did recently, for the update-motd package (from scratch), I used: [04:31] dh_installinit -- start 81 2 3 4 5 . stop 19 1 . [04:31] kirkland, so what does that actually translate into meaning? [04:31] superm1: 81 = start priority [04:32] ah and then 1 2 3 4 5 runlevels [04:32] 19 is stop priority, only stop in 1 runlevel [04:32] right [04:32] so I think you want the start priority at 24 [04:32] superm1: and the stop at 100 - 24 [04:32] superm1: 76 so says my math :-) [04:33] superm1: you want it to start in 2 3 4 5 run levels [04:33] well how to decide what runlevels? [04:33] superm1: and only stop in 1 [04:33] superm1: okay, so here's where my lack of upstart knowledge starts to show [04:33] isn't this doubled up though, the same things go in the header of the init script itself (which apparently is wrong too)? [04:33] superm1: and perhaps some of my long-time RH background.... [04:34] superm1: the stuff in the header of the init script is comments only, as far as I understand [04:34] superm1: and yes, those should be cleaned up as part of this patch [04:34] kirkland, oh, i wonder why lintian actually cares about that stuff then [04:34] kirkland, should there be any worry for upgraders with this though? [04:34] superm1: i think it's cleanliness and good manners [04:35] superm1: there could be more to it than that... i'll ask some people smarter than I tomorrow :-) [04:35] superm1: good question... i'd hope the debhelper would take care of that cleanly for you [04:35] kirkland, woah actually changing the top of the init script did change behavior (changing Default Start to 2 3 4 5) allows update-rc.d to not bail out [04:35] superm1: but I'm not sure [04:36] superm1: interesting.... i do agree that those need to be kept in sync and accurate [04:36] superm1: but I don't know who/what reads/cares about that commented data [04:36] kirkland, well perhaps i'll have to look at another service package like mysql and just try to emulate the way that are doing it. it's odd that it bailed out suddenly in intrepid [04:36] superm1: good call, mysql is well packaged [04:37] kirkland, yeah i definitely never agree with reading commented data, but empirical results seem to indicate it matters in this case [04:37] superm1: gotcha [04:37] kirkland, okay well thanks for the push in the right direction here :) [04:37] superm1: back to another question of yours.... you do understand the various runlevels, right? [04:37] superm1: no prob, not sure how much I helped, other than moral support ;-) [04:38] kirkland, i've got a basic understanding of them, but as upstart was introduced in the mix not as much [04:38] superm1: agreed, Upstart is the monkey-wrench [04:38] superm1: something like this might help just a bit http://www.debianadmin.com/debian-and-ubuntu-linux-run-levels.html [04:38] superm1: or even wikipedia [04:39] kirkland, ah thanks. i'll see if things appear to look more clear after looking over that [04:39] superm1: sure [04:40] superm1: in brief, setting it to run in 2 3 4 5 basically says "run myth-backend once the system is in any multi-user runlevel" [04:42] superm1: which does not include runlevel 1 (single user mode), which is primarily used for system rescue [04:42] kirkland, so why are 3-5 even around then? with upstart being in the mix, is graphical login really defined for level 5? [04:43] superm1: in the RH world, the only ones that make sense are 1 (single user), 3 (command line only, with networking), and 5 (full X stack, with networking) [04:44] superm1: and 2 and 4 are reserved (not used) [04:44] kirkland, so perhaps in the ubuntu world 3-5 are really just around for compatibility reasons then? [04:44] superm1: Upstart isn't quite read to supplant rc*.d yet [04:45] superm1: and thus the various runlevels are still around [04:45] superm1: exactly [04:45] kirkland, ah that seems to make more sense then [04:46] superm1: from the Upstart wikipedia article, "Easy transition and perfect backwards compatibility with sysvinit were explicit design goals" [04:48] kirkland, so looking at mysql-server's dh_installinit line, it look like dh_installinit -a --name=mysql -- defaults 19 21 is used [04:48] that would mean 19 is the default starting priority and 21 is the default stopping priority [04:48] with the runlevels determined by dh_installinit, correct? [04:49] superm1: yeah, look on your mythbackend: /etc/rc0.d/K21mysql [04:49] superm1: and /etc/rc5.d/S19mysql [04:49] superm1: remember, rc0 is the shutdown runlevel [04:50] kirkland, right, this seems sensible and easy enough to fix then [04:50] superm1: i think "S" stands for "start", and "K" for kill [04:50] superm1: also, please consider whether or not a shutdown action is necessary for mythbackend [04:50] superm1: there's been some effort to reduce the number of unnecessary shutdown scripts [04:50] you mean optionally just leaving the backend running? [04:51] superm1: as it slows down shutdown/reboot [04:51] superm1: well, just consider if its necessary [04:51] kirkland, dont most processes not like that kind of behavior and not having an option to clean up? [04:51] superm1: there exist some daemons that don't really actually need to be shutdown by it's init script (killing the process is clean enough) [04:51] kirkland, well i suppose i've never tried, but that should be something explored later i believe [04:51] superm1: most, right. but not all. [04:52] superm1: fair enough, just thought i'd mention it [04:52] kirkland, so if no shutdown scripts were included, how is that achieved? [04:52] not installing with defaults and two priorties, but instead just start and one priority? [04:53] superm1: i think that's right, but I'd have to test it out ;-) [04:53] superm1: we can look at that later, if you wish [04:53] kirkland, okay for now then defaults will suffice, and i'll try to remember to look at this later indeed :) [04:53] superm1: just something I thought I'd mention in passing [04:54] superm1: i have a feeling that mythbackend will want an explicit shutdown [04:54] yeah i seem to think it needs it because it has to issue commands to a master backend to say it's leaving the netwrok [04:54] superm1: right, that's probably enough right there [04:55] superm1: consider, for a moment, mythtv-status, though [04:55] superm1: the thing that updates /etc/motd.... [04:55] kirkland, yeah mythtv-status most definitely doesn't need it on shutdown at all [04:55] superm1: there's no good reason that needs to run on shutdown [04:55] superm1: right, so that's just your counterexample, in the interest of education ;-) [04:56] kirkland, well this type of exercise is indeed important and should be done on all types of daemons though if there is going to be an improvement [04:56] superm1: you bet. [04:56] superm1: were you willing to add the status action to the myth-backend init script? [04:56] superm1: i'd really, really like to have that one on my backend ;-) [04:57] superm1: i sent a patch at some point, you asked about backporting [04:57] kirkland, yeah i remember the issue was regarding older releases not supporting status [04:57] superm1: i can work something up that would work for hardy pretty easily, but the code would be slightly different [04:57] kirkland, how about this for a happy medium; [04:57] kirkland, install different init scripts in debian/rules depending on which release it is getting built against? [04:58] so there would be a new style init script for newer releases that support status [04:58] and once the earlier releases that didn't support it hit EOL, the old style one got dropped [04:58] superm1: hmm, if you're okay with the code duplication, i suppose that would solve the problem [04:59] superm1: i'd think you'd want to avoid two copies of nearly identical init scripts [04:59] kirkland, well the other option is to append the function to init scripts on releases that support it [05:00] superm1: right, so within the initscript, we could see if the status_of_proc() function is defined [05:00] superm1: and if not, then we'll define it [05:00] superm1: actually..... [05:00] superm1: it can be done a little simpler than that [05:00] kirkland, more or less. and once the earlier releases that don't support it are gone, drop that snippet [05:00] superm1: let me dig through some patches i've done for dapper [05:01] superm1: some people asked me for help getting something similar working for dapper [05:01] superm1: and i put something together [05:01] superm1: how about i send you a patch tomorrow for your review [05:01] kirkland, sure, if you want to drop me a patch sometime tonight, i'll throw it in the upload i'll do. [05:01] superm1: with the explicit goal of having a single init script that works for Hardy & Intrepid [05:01] kirkland, or tomorrow should be fine, i suppose it is getting late [05:01] superm1: yeah, i need to call it a night [05:02] kirkland, i'll put off the upload pending hearing back from you then [05:02] superm1: i'll have a much easier time with this tomorrow, when i'm fresh, and have some coffee ;-) [05:02] superm1: okay, i'll do it first thing, won't take very long at all [05:02] kirkland, okay just give superm1|away a diff then. have a good evening :) [05:02] superm1: take care of that ankle ;-) [05:03] thanks [05:05] what's up MOTU land [05:05] have a few minutes to bug ya while my smoke tests complete so I can go to bed [05:06] nixternal! [05:06] * nixternal needs to get list admin stuff on his worky lappy or on the server so he doesn't have to fire up another machine [05:06] wasabi ajmitch [05:06] my job is trying to kill me I think [05:06] * ajmitch is just amusing himself watching people 'debate' on IRC about stuff [05:07] note to self: do not continuously press ctrl+o in mutt thinking it is ctrl+p instead [05:07] heh [05:08] you can only open a folder so many times [05:08] nenolod: you should avoid feeding trolls, btw :) [05:09] ajmitch: lol. [05:09] ajmitch: but he is so amusing [05:09] sure... [05:09] how long ago did this happen? I like watching and reading about trolls [05:10] trolls == success [05:10] #chromium-linux [05:10] ahh [05:10] ongoing discussion [05:10] tell the troll to package RPMs all day, I bet they quit trolling [05:11] sounds like fun! [05:11] hell no, I hate it! [05:11] * ajmitch wishes he got to package RPMs all day long [05:11] you can have my job if they don't change to at least Ubuntu/Debian soon [05:12] heh [05:12] I am trying to get them to work on a partnership with rPath now [05:12] if they don't want to do that, then I will talk with Mark about a partnership [05:13] they don't want to pay for rMirror, but rMirror and rBuilder cannot be beat in the appliance world [05:13] * ajmitch doesn't know who you work for or what you package [05:13] I work for a company called Cleversafe (http://www.cleversafe.com) [05:14] we do distributed storage appliances with a goal of eliminating Red Hat and Amazon in their clouds which are useless [05:14] interesting [05:14] I am tired of looking at Python, Java, and RPMs [05:14] but python is fun [05:14] I need another C++ job [05:15] * ajmitch mostly does PHP at work, so it can't be all bad for you [05:15] perl is more efficient, no matter how much I hate writing in it [05:15] * RoAkSoAx does nothing at all :P xD [05:15] right now udev is kicking my arse with a "fake arse hot-swap" in appliance regeneration [05:16] amazon EC2 is way expensive [05:16] I have the pulls working just fine...I even have the inserts working just fine and my killer Python script regenerating the appliance...but when I start up the machine, my udev rule gets called and creates a groovy kernel panic [05:17] EC2 is inefficient imho, especially considering the storage options are sans [05:17] nixternal: i am trying to bring rPath/conary-style inheritance to debian. but probably this will become a lenny+1 thing. [05:18] lenny+1 isn't far off [05:18] with cleversafe we have what is called a dsnet....you have a minimum of 8 storage machines (slicestors) that break up your data into very small encrypted bits (the only proprietary part of our software btw) and stores them in various locations (vaults) on the dsnet [05:19] ajmitch: i have it working in my own playground distro [05:19] nenolod: which functionality? rollback I am assuming [05:19] nixternal: shadow packages, and policies. copy-on-write, and rollback would have to be implemented into Dpkg. [05:19] migration is an appliance developers best friend [05:19] yay shadow pkgs [05:20] rollback wouldn't be terribly hard, especially if COW was working [05:20] dpkg is a lot more resilient to faults than RPM as it is, so i guess it is not seen as important [05:20] but it would be cool ;) [05:21] rollback would be much better if dpkg only applied changed files [05:21] instead of complete binaries [05:21] well, sadly that would require querying a server [05:21] probably not going to happen :P [05:22] or some obnoxious package format resembling xdelta on crack [05:22] oh wait, that's what conary changesets are :P [05:22] hahhaa, I was just gonna bring up xml/xdelta/and some other stuff [05:22] * ajmitch hasn't really followed what is different & special about that stuff [05:22] i played with foresight and found conary to be quite inspiring [05:23] however, conary does a few things wrong, like depending on external servers to find out what deltas to pull [05:23] ajmitch: conary & rPath is appliance centric, whereas every other os is server/desktop centric [05:23] nixternal: nice sound bite, now explain it :) [05:23] Ubuntu is trying with JeOS, but they have to get rid of that crap kernel and come up with a fully automated kernel editor/config/builder [05:23] nixternal: i think some concepts from conary could be applied to dpkg and apt [05:23] nenolod: +100 [05:24] however, debian will never be a good platform for appliances unless debian-installer gets a total overhaul [05:24] well, I will disagree on that....d-i with FAI totally rocks [05:25] but d-i with pxe and kickstart can be a pita [05:25] well, the way i had in mind is quite simple [05:25] I should say kickstart is the d-i pita there [05:25] Also, vm-builder does a nice job for virtual applicance [05:25] rBuilder ftw :) [05:25] basically, you have your base system repo [05:25] and then you have override repos [05:25] which provide policy packages, shadow packages, etc [05:26] I can select iso for app or vm, or I can select a vmware/vbox/qemu image from a drop down [05:26] if it were done this way, ubuntu could reduce the diff between debian [05:26] same thing I deal with utilizing CentOS for an appliance [05:27] I have a super nasty rsync script that will check for changes, and move the changes to a temp file until they can be researched, tested, and then included [05:29] * nixternal kicks maven in the arse and goes back to ant [05:35] nixternal: See https://wiki.ubuntu.com/JavaTeam/Specs/MavenSupportSpec about maven: it's not suitable yet (although if you'd like to work on it...) [05:36] haha, nevah! [05:36] it sucks in Fedora, Red Hat, CentOS and * [05:36] once it is mature and a hell of a lot less confusing, it will be good stuff [05:36] but ant just works and works easily to be honest [07:14] ScottK: ping [07:17] persia, poke [07:18] NCommander: I'm just heading off: please be quick (and please provide content next time, or I may ignore the poke) [07:18] ScottK: when you have some time would you take a look at bug #263405 for me please? [07:18] It's an FFe for ampache. [07:18] Launchpad bug 263405 in ampache "update Ampache-3.4.1 to Ampache-3.4.3" [Undecided,New] https://launchpad.net/bugs/263405 [07:18] persia, if I provide a contextless poke, it usually means I just want to chat, no specific note to leave ;-) [07:19] persia, so cya ;-) [07:20] ScottK: If I give a slow response it's probably because I fell asleep in my chair, it's quite late for me =) [07:20] kirkland, we're rolling some test disks tonight and that bug was keeping them from rolling. i uploaded myth tonight, and will upload again after I see your patch okay? [07:36] good morning === superm1 is now known as superm1|away [08:07] Good morning Daniel! [08:23] hey iulian [08:32] Im looking at the net pidgin 2.5.1 and there is a debian/libpurple0.symbols file - how do i check this is still correct? anyone able to help me understand symbols or can point me to some documentation on understanding them? [08:33] I have this - https://wiki.ubuntu.com/PackagingGuide/Recipes/CheckingLibrarySymbols (not sure that this is entirely the same thing thou) [08:38] stefanlsd: ah packaging pidgin 2.5.1 for intrepid? ;) [08:38] nodnod [09:08] morning everyone [09:13] stefanlsd: cool ( im a subscriber @ bugreport, that's why i asked :) ) [10:19] LP:263074 should this one be synced to intrepid or is it too late? [10:22] Coper: it'll need a FFe, but that sounds like a good candidate to get synced. [10:22] * Hobbsee notes people like ScottK might just poke it through [10:32] isnt there some launchpad firefox plugin? [10:32] for the search bar? [10:32] so you can just type in the number in be directed the the launchpad page [10:34] there is for debian. not sure on ubuntu [10:34] you could create one, or just create a bookmark that works. [10:34] ie, lbp expanding to the correct bug. [10:36] is "smart bookmark" the appropriate term in firefox? (that's what the similar thing in epiphany is called, at any rate) [10:53] there, made one myself :) [10:53] somebody wants it? :) [11:03] Hobbsee: okey so I can create a debdiff for that package and submit to the bug? [11:04] Coper: why would you be creating a debdiff, if it was a sync? [11:04] (or is that the new procedure for ffe's now?) [11:05] I don't have the process clear to me how a sync is working.. How is making a sync? [11:05] Who* [11:12] Coper: Your two options for getting a package from Debian into Ubuntu is - Sync or a Merge [11:13] For both of these, you file a LP bug requesting a Sync or a Merge [11:13] you need to determine if what you are trying to do is a sync or a merge [11:13] A sync will be if the debian package will work exactly as is in Ubuntu and no Ubuntu changes are required in the package. [11:14] We see this by packages in Ubuntu that are package-ver-1 for example. [11:14] If we see a package in Ubuntu package-ver-1ubuntu1 we know there are Ubuntu changes. [11:14] okey so in this case it's just a sync becurse we dont have any ubuntu versions of it. [11:15] When then have to decide, are the changes that Ubuntu did specifically to the package still necessary to keep it an Ubuntu version. If they are, we need to merge. Else we file a sync bug. [11:15] Ideally, any changes we make to Debian packages to make an ubuntu1 version, are given back to Debian, so next time we sync or merge, we can sync. We always want to try stay as much as possible in line with Debian, else we just make lots of work for ourselves [11:16] Coper: Yeah, if it was originally a Debian version and we made no changes, we will sync. Being in FF at the moment, we need to make sure the sync fixes bugs and doesnt introduce new features. [11:16] of course, Ubuntu Main can complicate the whole sync thing. but for universe, go for it if you can [11:17] If that is the case, you can apply for a Freeze Exception - https://wiki.ubuntu.com/FreezeExceptionProcess === dholbach_ is now known as dholbach [11:29] devfil: ping me when you will need a sponsor for #490586 [11:30] POX_: #490586 ? [11:30] http://bugs.debian.org/490586 [11:31] POX_: ah ok, it is a sort of alpha for now, I'm awaiting for a release more stable [11:31] ok [11:32] s/awaiting/waiting/ [12:32] should sync requests be subscribed to anyone special? [12:33] torkel: I'd suggest the sponsoring queue, if you are not a developer [12:33] nope. I'm not. [12:34] torkel: Using "requestsync" (in intrepid) shall automatically subscribe team for you. [12:35] the necessary team* [12:36] jpds: I used lp to file a bug (#264303) :-( [12:36] torkel: you will probably need and FFe also https://wiki.ubuntu.com/FreezeExceptionProcess [12:37] torkel: Looks good, might want to use the script next time tho. [12:37] porthose: why? It's not a new upstream version, just a new debian revision [12:38] ahhh ok [12:39] jpds: yeah. There are too many fancy tools to remeber them all though. Especially when you only do random requests [12:41] Hi, I cannot install python-dev on 8.04 because of unmet dependencies: http://paste.ubuntu.com/43010/ [12:41] it's ubuntu-universe-sponsors I should subscribe? Right? [12:42] torkel: is the package in universe? [12:42] yes [12:42] balachmar: any chance the mirror is not in sync yet [12:42] torkel: then yu are right [12:43] slytherin: thanks :-) [12:44] bug #264303 [12:44] Launchpad bug 264303 in openafs "Please sync openafs (1.4.7.dfsg1-5) from debian" [Undecided,New] https://launchpad.net/bugs/264303 [12:44] torkel: No need for a freeze exception for that sync. [12:44] slytherin: Well I selected the main server and it doesn't install [12:45] balachmar: I don't have a hardy installation right now, so can't help much [12:46] persia: do you know anything about symbols? [12:46] stefanlsd: Some. What do you need to know? [12:47] persia: Im trying to understand why they are needed. Is it for debug purposes? [12:48] stefanlsd: Please define "symbols", as there are many possible answers to your question, depending on what you mean. [12:48] slytherin: too bad. I'll check my sources.list again [12:49] persia: im looking at pidgin 2.5.1 and we include debian/libpurple0.symbols. dh runs dpkg-gensymbols and makes a diff (not sure sure between what exactly, or how dpkg-gensymbols gets them). Im trying to work out, why we even need these symbols, as its not in every package [12:50] persia: the package builds and im running it fine, im just trying to work out what i'm missing... [12:50] stefanlsd: OK. Those are the library symbols. [12:51] Essentially, when creating a library, we need to have a set of names (symbols) to call each of the exported functions, and we need to let programs that use the library use the same symbols to identify what they want the library to do. [12:51] I have turned of all extra repos and now I can select force version, but it doesn't force the version! ... [12:52] The python version was from hardy-proposed and has been pulled from hardy proposed (As I am told) [12:53] persia: ok. and the names may change in the library. like libpurple0 may have a different symbol now for a function? [12:53] But now it doesn't want to go to the original hardy-security version [12:53] dpkg-gensymbols hunts down libraries, and creates a debian/symbols file : this file is then used to better identify the versioned dependencies of a program using the library (from ${shlibs:depends} [12:53] stefanlsd: Maybe. Depends on how much the code has changed since it was last compiled. [12:54] balachmar, that's your problem [12:54] persia: ok. that makes sense. why do some programs use it, and some not? Or will all libs use package.symbols? [12:54] balachmar, using -proposed is fraught with risk. [12:54] stefanlsd: All libraries that export symbols (and a .so file) ought use it. [12:55] balachmar, look at the entire list of python-dev dependencies that it's moaning about, and use "aptitude reinstall fooapp=2.5.2-2ubuntu4.1" until all remnants of -proposed are gone [12:55] It's mostly important for C and C++ programs, although some other languages also use this library format sometimes. [12:55] persia: https://wiki.ubuntu.com/PackagingGuide/Recipes/CheckingLibrarySymbols - is this still current. Or is this now obselete with the introduction of dpkg-gensymbols. and why cant we just use what dpkg-gensymbols finds for us? isnt that its purpose? [12:56] stefanlsd: That page becomes obsolete when every package providing a library uses dpkg-gensymbols, which I don't believe to be the case today. [12:57] That said, it's worth doing some research and updating the page with a note about the use of the new system. [12:57] persia: hehe. i would love to as soon as i get to the bottom of this :) [12:58] directhex: I didn't know that is was that risky. Read somewhere that is was a good way to help out. With bug finding. But I wasn't aware that I could get into dependency traps... [12:58] balachmar, it IS a good way to help with bug testing... until someone finds one, and pulls a package from there [12:58] balachmar, then you're stuffed really [12:59] or at least need manual intervention [12:59] directhex: As I am aware of now... :) [13:00] persia: http://pastebin.ubuntu.com/43037/ - thats what dpkg-gensymbols wants to do. [13:03] persia: does this mean i should be editing my debian/libpurple0.symbols accordingly? Is dpkg-gensymbols to be trusted? :) [13:04] PS: A nice tip i got was to export DPKG_GENSYMBOLS_CHECK_LEVEL=4 which causes pbuilder to not complete if the symbols are broken :) [13:13] directhex: Thanks for the help! I fixed it (and I will leave proposed off on my work machines..) [13:28] stefanlsd: This means you should follow the advise on the wiki page, and make sure that you have the right symbols file. I've not used dpkg-gensymbols myself, but suspect it may be correct, although if you're not sure of anything, it's always best to double check. [13:28] persia: oki. thanks persia. that helped a lot. If i can work it out, i will update the wiki page. === asac_ is now known as asac [13:56] is there some reason why DEB_INSTALL_DOCS_ALL works but DEB_INSTALL_EXAMPLES_ALL not? [14:00] can anyone explain me why this is not a bug - bug #264283. I fail to understand Mr. Pedro's understanding. [14:00] Launchpad bug 264283 in rhythmbox "Ripping preferences have no 'Strip Special Characters'" [Low,Invalid] https://launchpad.net/bugs/264283 [14:00] RainCT: Well, I can probably explain why DEB_INSTALL_DOCS_ALL works, but not so much why DEB_INSTALL_EXAMPLES_ALL doesn't. [14:01] slytherin, because in gnome land, you get one tick box and a thousand regist^Wgconf settings [14:01] AND YOU LIKE IT! [14:02] slytherin: Ask pedro? [14:02] directhex: I wouldn't mind it if it was the case with sound-juicer also. But sound juicer had a visible option, and rhythmbox does not have it. [14:02] soren: will do the same. [14:03] RainCT: DEB_INSTALL_DOCS_ALL is needed because cdbs has some files that it by default puts into each package (i.e. some files that it always wants to pass to dh_installdocs). It has no such thing for dh_installexamples. [14:03] slytherin, patches accepted! [14:03]