[00:10] xnox, I've thought about trying to get debian to move to hdf5-1.8.3 but they say they have their reason to stay on 1.6.6.... so the problems I'm having is I can get the hdf5 deb to build and it installs the files in the right places, but then when I try to build packages that depend on HDF5 i get the following error: dpkg-shlibdeps: failure: no dependency information found for /usr/lib/libhdf5_cpp.so.6 [00:10] I've checked in /usr/lib/ and the libhdf5_cpp.so.6 is there but for some reason its not found when trying to build debs that depend on it [00:20] so I guess I was looking to see if someone could coach me through this problem [00:46] neurobuntu_: That error is not saying that the library itself could not be found - rather that the debian dependency information that enables dpkg to build proper package dependencies is missing [00:47] neurobuntu_: Normally this file would be constructed by dh_makeshlibs during the hdf5 deb build process [01:07] maco, endless thread is endless [01:08] directhex: s/endless/pointless/ [01:09] maxb, so I call dh_makeshlibs during build, is there a way that i can verify that dpkg is building the proper dependencies? [01:09] I call dh_makeshlibs during the build of hdf5 [01:09] What do you mean by "proper"? [01:10] There should be a shlibs control file present in the final .deb [01:10] ajmitch, nah, at some point mark will force an epiphany from those who disagree with him, and shuttleworth himself will descend in his space ship to explode the offending packages with lasers [01:11] ಥ_ಥ ...Beautiful. [01:14] directhex: that thread totally went plonk [01:14] maxb, thanks for your help the control file isn't in there [01:16] dtchen, went? [01:17] directhex: killfiled the mono thread [01:18] dtchen, /dev/null seems somehow insufficient. can you pipe into shred? [01:19] sorry, already breaking other ubuntu components with shred ;-) [01:20] dtchen: Like pulse? :-P [01:21] :-P === asac_ is now known as asac [01:25] neurobuntu_: Is your source package published anywhere? [01:26] maxb, yes its up on launchpad [01:26] maxb, https://launchpad.net/~soma-testing/+archive/ppa [01:35] so my rules file does have the command: dh_makeshlibs -phdf5-serial-1.8.3-0 -v hdf5-serial-1.8.3-0 but when i examine the contents of the compiled deb with dpkg -c I don't see a control file [01:36] this is what i get http://paste2.org/p/253075 [01:48] directhex: so, starting on the port/rewrite of f-spot to C? [01:49] ajmitch, i heard hub was thinking of doing it [01:49] oh good [01:49] ajmitch, let's ignore the fact that his implementation method involves implementing parts of ECMA 334 [01:49] but it's not on the mono runtime, so it's ok [01:49] if only pnet didn't suck [01:50] it wouldn't make any difference what the runtime was called [02:01] neurobuntu_: The shlibs file is present - the significance of it being a "control" file is that it is not part of the data files shown with -c [02:02] 'dpkg -I foo.deb' lists the control overview info, 'dpkg -I foo.deb shlibs' shows the shlibs control file specifically [02:02] However, the content of it looks a bit wrong to me [02:04] maxb, what kind of things look wrong? [02:05] The version field of the shlibs line is wrong [02:06] * h6w wonders how others do package management from cvs/svn/bzr. [02:06] I'm using ant (command-line). If I create a release folder and ignore it from svn/bzr then put the src zips, previous packages, and diffs in there, does that make sense? [02:06] I'm replicating this process across a number of projects, so I want to get it right for most/all cases. [02:07] So, I was thinking of having an ant script that looks for that folder grabs the most recent package/diff in there for that stream (e.g. Nightly), takes a release version parameter, then creates a new package/diff from the current src and puts it back in the release folder. [02:07] neurobuntu_: In 'dh_makeshlibs -phdf5-serial-1.8.3-0 -v hdf5-serial-1.8.3-0' what is the second repetition of the package name trying to achieve? I think it may be redundant / broken [02:07] maxb, I'm not sure... I was using the debian dir from hdf5-1.6.5 as a template and that is how they did it [02:09] i just recompiled to see the lintian errors, some of them might be relevant http://paste2.org/p/253096 [02:15] It's unrelated to the shlibs problem, but your ppa upload is accidentally a native package, btw [02:15] .tar.gz, not .orig.tar.gz and .diff.gz [02:17] maxb, i'm aware... [02:22] Gah, what on earth have upstream done with their tarball!? [02:22] Somehow they've got two entries in the tarball with the same file name! [02:24] Anyway [02:25] neurobuntu_: So the problem with the shlibs is that they have an explicitly built shlibs file in the package, which doesn't correspond with the libraries actually built in this new version [02:25] in upstream? [02:25] See the shlibs.in file and the code which processes it into a shlibs file in debian/rules [02:25] No, in the previous debian package [02:26] ahh... [02:29] maxb, they use that @VERSION@ variable all over the place and it has been giving me problems in the past... do you think i should just replace it with 1.8.3 [02:29] if anyone has some time to revu http://revu.ubuntuwire.com/p/pidgin-mbpurple , I would be greatly appreciative :) [02:30] neurobuntu_: No, that's not the problem - it's apparently expecting the version number to be present in the library filenames, but in your case: the version is NOT in the name [02:31] huh? i'm confused how is the version number not in the filenames? [02:31] :-/ [02:33] isn't 1.8.3 the version number? [02:35] Yes, and it's not in your library filenames [02:36] maxb, you're talking about the names of the .debs created or something else? [02:36] It looks like you've not forward-ported the debian customizations to the build system [02:36] I'm talking about the names of the .so files [02:36] hmm... ok [02:37] You're missing debian-specific changes to the libtool invocations, which is why your build results don't agree with what the package is expecting [02:37] maxb, might it be better to build this on a debian machine then? [02:37] except that won't gaurentee it works with ubuntu then.... [02:39] Uh, no, that's not what I mean at all [02:39] I mean debian-specific as in "not upstream" rather than "not ubuntu" [02:40] maxb what would you recommend I do? try to get someone who knows more then me to package it for me.... start from square one and get rid off the debian/ i borrowed from 1.6.5? buy out a motu for a weekend? [02:40] the more I learn about library packaging the more I realize i don't understand and I'm really starting to doubt I can get this thing package properly [02:41] The problem is that you've taken the Debian hdf5 1.6.x package as a base but you have not properly updated the patches it applies to the upstream source to apply to 1.8.x [02:42] Unfortunately your job is made much harder by the way they've apparently completely changed the buildsystem upstream [02:45] :-| [02:48] I don't suppose debian intend to update their package any time soon? [02:50] maxb, nope I was asking to see when they were going to update and nobody could tell me [02:51] when you say i "need to update the patches" you're not referring to patches in the debian dir are you? when i do apt-get source libhdf5-serial-1.6.6 I don't see any patches under the debian dir [02:52] zcat hdf5_1.6.6-4ubuntu1.diff.gz | lsdiff [02:53] The .diff.gz doesn't just create the debian/ dir, it also changes some of the upstream files [02:54] Before you get too scared looking at that... some of the files changed are the result of autogeneration [02:54] but some are not [02:58] maxb, wow ok well I can't express how helpful you've been tonight I'll hit the net and try to better understand what debian is doing before they package.... thanks again for your help I'm sure i'll end up in here again sometime [03:00] neurobuntu_: you've seen the README.Debian_shlib file, right? [03:02] yes [03:02] The key additional libtool flags regarding that change are -release $(H5_VERSION) -version-info 0 [03:05] so does that readme mean that i have to setup the appropriate symlinks [03:06] what is this repos? [03:06] http://archive.ubuntu.com/ubuntu/pool/main/ [03:07] jaunty or karmic? [03:07] Every release currently on archive.u.c [03:07] Dapper, Hardy, Intrepid, Jaunty and Karmic [03:09] my local repos created from the CD only contains one single file as: [03:09] ubuntu/pool/main/r/ruby1.8/ [03:09] libruby1.8_1.8.7.72-3_i386.deb [03:09] why? [03:10] StevenK: ^ [03:10] I have no idea, since I don't know how you created your local repo [03:11] maxb, thanks again for your help [03:14] qiyong, is its a repo from a liveCD? [03:14] neurobuntu: yes, downloaded from the site [03:15] a full cd thould be [03:15] Oh, a live CDs repo doesn't contain a lot [03:15] It's pointless to use that as a base [03:16] qiyong, if you are using a distribution before jaunty check out falcon [03:16] if you want to build your own repository [03:16] or launchpad [03:17] should I use apt-mirror to build my local repos? [03:18] apt-mirror works, I use it for my local mirror [03:19] http://releases.ubuntu.com/9.04/ubuntu-9.04-server-i386.iso.torrent [03:19] is this the said live CD? neurobuntu [03:19] No, that's a torrent for a sserver install ISO [03:19] s/ss/s/ [03:20] i mean the iso downloaded by that torrent [03:20] is it the live CD? [03:20] i thought it should be a full CD [03:21] StevenK: ^ [03:21] * DBO pokes directhex [03:21] What do you mean by "full CD" ? [03:21] contains all the pkg from jaunty [03:21] No such CD or DVD exists [03:22] Since Jaunty is 18GB for i386 [03:22] or binary main at least [03:22] oh [03:22] Binary main is still too large for a DVD [03:22] how much is a binary main? [03:22] For Jaunty? [03:23] yeah [03:24] steven@infected:/srv/mirror/ubuntu/dists/jaunty/main/binary-i386$ echo $(($(grep '^Size' Packages | cut -d\ -f2 | numsum) / (1024 * 1024))) [03:24] 5787 [03:24] A shade under 6GB, if I can do math [03:24] a DVD can be 6G? [03:24] qiyong, dual layer can [03:25] We don't currently ship any dual layer DVDs on cdimage.ubuntu.com [03:25] is there that a full DVD? [03:25] full DVD iso? [03:25] oh [03:26] And full Jaunty i386 is 24GB [03:26] qiyong: Just use apt-mirror to grab what you need, instead of starting with an ISO. [03:26] i have this folders: [03:26] Yes [03:26] [DIR] dists/ 23-Apr-2009 15:16 - [03:26] [DIR] indices/ 09-Jun-2009 00:57 - [03:26] [ ] ls-lR.gz 09-Jun-2009 00:56 7.1M [03:26] [DIR] pool/ 14-Jan-2008 22:05 - [03:26] [DIR] project/ 13-Feb-2008 14:39 - [03:26] to be a mirror or repos, which folders are must? [03:26] pool? [03:26] pool and dists [03:27] But seriously, use apt-mirror [03:27] qiyong: See http://popey.com/Creating_an_Ubuntu_repository_mirror_with_apt-mirror for one tutorial on doing this. [03:39] $ apt-mirror [03:39] Downloading 84 index files using 20 threads... [03:39] apt-mirror: can't write to intermediate file (index-urls.0) at /usr/bin/apt-mirror line 174. [03:40] ignore above [03:44] /var/spool/apt-mirror/mirror [03:44] /var/spool/apt-mirror/skel [03:44] /var/spool/apt-mirror/var [03:44] what these three dirs for indivadually? [03:45] can I only mirror one release only? mirror only jaunty for instance [03:46] i'm sure you an [03:46] *can [03:47] what about the three folders? [03:47] i don't know. Does the manual or hte howto tell you? [03:55] Hobbsee: check later [03:55] ? [03:56] how do I find who is using a mounted fs? lsof or fuser which what options? [03:57] fuser -m [03:57] * Hobbsee advises #ubuntu for non-packaging questions [04:07] ajmitch: [04:07] root@leo:~# fuser -m /dev/sdb [04:07] root@leo:~# umount /srv/sdb [04:07] umount: /srv/sdb: device is busy. [04:09] try calling that on a partition, not on an entire drive. [04:10] if anyone has some time to revu http://revu.ubuntuwire.com/p/pidgin-mbpurple , I would be greatly appreciative :) [04:39] Hobbsee: i have /dev/sdb mounted on /srv/sdb [04:40] qiyong: Yes, and that works for you? If so, you must have format the disk as a partition. This tends only to happen on SSDs. [04:41] Normally, you'd have a partition number at the end. e.g. /dev/sdb2 [04:44] umount: /srv/sdb: device is busy. [04:44] h6w: ^ how do I find who is using a mounted fs? lsof or fuser which what options? [04:44] i need to umount it [04:44] Are you in that dir? [04:46] no [04:47] fuser -c -u /srv/sdb [04:47] nothing [04:48] Then nothing is using it. [04:48] i want to find what process using sdb [04:48] Try it with sudo [04:48] still umount fails [04:48] i'm root [04:48] Well then it must not be umounting for some other reason. [04:48] What type of fs is on it? [04:49] ext3 [04:50] You could use umount -l or umount -f, but you need to be sure everything's written to it. [04:54] h6w: ok, -l get it umounted [04:55] cool. man is your friend. [07:11] good morning [07:13] 'ey Daniel. [07:17] hiya iulian [07:17] how's life? [07:24] dholbach: Everything is good :) [07:24] dholbach: How about yours? [07:25] very good - just going through the sponsoring queue right now [07:25] lots of good stuff in there [07:26] Indeed, I was taking a peek at it yesterday. I'll dive in more today. [07:27] * dholbach hugs iulian [07:27] you ROCK [07:28] * iulian has to study for his driver's license exam :( [07:28] * iulian hugs dholbach back. [07:39] bdrung_: I need to install 246724967246MB of build depends just to build the matplotlib source package! :-) [07:39] ah, but is one of them yada? [07:40] no, luckily not [07:42] dholbach: I can build it for you. [07:42] * iulian has a fast net connection. [07:42] iulian: no worries - I have them installed now [07:43] OK, great. [07:45] 246724967246 MB at 1Gbit/second would take around 62 years... iulian must have a *very fast* net connection :) :) [07:46] jmarsden: we're not dealing with trifles here, we're changing the world! :-) [07:46] As long as it doesn't delay the release or Karmic by 62 years, that's fine with me :) [07:48] Heh [07:49] jmarsden: if dholbach has enough free space to store the data I guess he has also a fast enough connection to receive it [07:49] geser: Impressive :) [08:08] mterry: not sure if you heard of "submittodebian" before, it makes use of https://wiki.ubuntu.com/Debian/Usertagging (really good stuff) [08:12] morning o/ [08:26] <\sh> moins [08:32] <\sh> bah...the trolling didn't stop... [08:33] which trolling? [08:33] \sh: server side filtering for the win. [08:33] hyperair: the mess on u-d-d list [08:33] oh the mono one? [08:33] * hyperair facepalms [08:33] <\sh> Hobbsee: thinking about that, unsubscribe u-d-d looks like a better plan [08:34] i wonder how old the troll is.. [08:34] \sh: indeed. [08:34] sounds like a kid making noise. [08:34] \sh: depends if there's enough useful stuff on it [08:35] <\sh> Hobbsee: tbh, I would like to see technical discussions more on u-d then discussing things with mostly-nonrelated-nontechnical-userbase could be that I'm too arrogant [08:36] \sh: if i'm figuring out what you mean right (your english grammar is stuffed there, so i'm guessing after the third read), that's what's supposed to happen [08:36] however, one of those "discussions" appears to be mono [08:39] <\sh> Hobbsee: the discussion is totally non-technical and not relevant. It's more godwins law with s/nazi/ms+boycottnovell/...actually it's only wasting bandwidth and mail-storage [08:41] \sh: oh, i know. If it was kubuntu-users, i'd moderate anyone who posted on the topic, send a warning mail that the discussion was to stop right here, right now, warn that i'd moderate the *entire* list if it didn't, and regularly go through the moderation queue. [08:41] unfortunately, it's not a list I have access over [08:41] oh, and there would be more in the mail than that (like, about the CoC), but that would be the crux of it. [08:43] <\sh> moins siretart` [08:43] gah, the other thread is no better either. [08:43] hey folks! hi \sh! [08:44] * hyperair is thankful that he is not subscribed to u-d-d =\ [08:45] <\sh> siretart`: I fixed the menu.lst problem like "write my own update-grub in scripts/GRUB/10-setup" ;) [08:45] <\sh> siretart`: and fai-divert update-grub [08:45] \sh: yes, but this breaks the purpose of ucf as well [08:46] <\sh> siretart`: problem I saw was that the menu.lst inside /var/run/grub/ wasn't correctly set...the name was menu.lst.ucf-dist and I wonder where that came from [08:46] \sh: BTW, are you targeting karmic or jaunty currently? [08:46] <\sh> siretart`: karmic [08:47] <\sh> aeh sorry jaunty [08:47] aha [08:47] <\sh> karmic just got grub2 ,-) [08:47] exactly [08:47] * \sh is still tired...1:30 UTC+2 I was at home...even if I wanted to go home early [08:48] * siretart` is going to have a long day today as well :-( [08:50] <\sh> siretart`: too much work here right now...we got 2x hp bl7000c full featured enclosures with complete 32x bl465c and 2x Cisco 6509 Carrier Grade 21U switches + 2x Cisco ASR routers...and we need to install them next week in our new DC...so the FAI stuff is for the blades ;) [08:50] I see [08:51] I'm looking forward to see a working FAI in ubuntu again :-) [08:51] <\sh> it's really no fun...even if I would love to play with the blades..but there is always no time :) [08:52] <\sh> siretart`: yes...debian lenny + this firmware not included crap is giving me headaches...when you install dl365 e.g. and you don't have a working NIC because broadcoms firmware is not included by default..no fun [08:53] ah, right, but isn't the firmware packaged in non-free? [08:53] <\sh> siretart`: but it looks like that SID kernel and jaunty kernel have problems mounting aufs over nfs, people on #fai@oftc reported that...I need to look at this later on [08:53] or is that the crappy sort of firmware? [08:54] <\sh> siretart`: yeah...but not installed by default because of non-free so you when you are not using d-i you need to take care to install it manually [08:55] <\sh> oh my english this morning is kauderwelsch [08:55] * \sh needs another tea [08:56] * xnox set up a filter to delete threads with "mono" in it from my mailbox [08:56] but with fai that should be rather easy to implement, or do I miss something important here? [08:56] xnox: try scoring [08:56] <\sh> siretart`: nope...you need to install it into the nfsroot [08:56] the most difficult thing seems to remember to do that :-) [08:57] * xnox uses gmail...... [08:57] <\sh> siretart`: but using ubuntu all the time, you don't expect such "feature" from debian...and hp supports debian officially so I thought it should run out of the box [08:57] perhaps a switch --with-nonfree-firmware can be added to make-fai-nfsroot or something [08:57] <\sh> siretart`: sounds like a nice wishlist bug ;) [09:21] <\sh> siretart`: jaunty is installed via fai ;) well..without update-grub and ucf but installed ;) === NCommander is now known as ApportRetracerPo === ApportRetracerPo is now known as NCommander [09:22] \sh: so aufs finally works again on nfs in jaunty? [09:22] s/again// [09:24] frankly i never once managed to get aufs to do anything. the manual is obtuse at best [09:26] <\sh> siretart`: no it's debian lenny with fai from uni-koeln.de installing jaunty [09:26] <\sh> siretart`: that's the "fast and furious installation plan"...now I can work on "fai on jaunty" [09:26] oh, I see :-( [09:27] ah, I see [09:36] jpds: the buildd script in u-d-t trunk has now full support of LP API for give-backs and rescoring [09:37] geser: Awesome! :D [09:37] dholbach: Then you need only 172473253 1,5 TB hard drives to store the data. :D [09:38] bdrung_: no big deal... I sponsored it - good work! [09:39] dholbach: thanks [09:39] geser: I guess this means we can finally remove lp_cookie from the source. [09:39] bdrung_: hey there. would you mind updating the mplayer backport in the motumedia PPA? [09:40] jpds: if it isn't used anywhere anymore sure (didn't check) [09:40] bdrung_: I've uploaded an upgrade path fix [09:41] siretart: i will do it this afternoon [09:42] jpds: from a look at some of those lp_functions methods, I guess they can use some caching to not refetch the same objects through the LP API everytime just to access an attribute (but I didn't have a close look yet) [09:43] geser: I'll take a look. [09:43] bdrung_: thanks [09:44] siretart: there is no comment in the changelog about a path fix [09:44] jpds: requestsync:import ubuntutools.lp.cookie as lp_cookie [09:44] bdrung_: upgrade path, not file path [09:46] geser: launchpad_cookie isn't used anywhere tho... [09:51] siretart: can you update the debian maintainer field the next time? dpkg-source complains: Version number suggests Ubuntu changes, but Maintainer: does not have Ubuntu address [09:56] bdrung_: the ubuntu package *is* actively maintained in the pkg-multimedia git repository in the ubuntu branch [09:56] that's why I've kept the maintainer field, because it is acurate [09:57] siretart`: it will break dpkg-buildpackage for people with a ubuntu.com mail address IIRC [09:59] dholbach: are you really sure with that? IIRC there are also some packages with a dedicated maintainer field in the archive, and I don't remember some mail that they were prohibited [09:59] i only ever remember it throwing a warning before [10:00] it's become mandatory now? [10:00] Hi, (newbie question) if I use dch -i on a "native debian" package of mine, and the previous changelog entry is "ltsp-server-pnp (0.1.4) jaunty; urgency=low", I get a proposed new version of "0.1.4ubuntu1". How can I make dch use "0.1.5" as the new version? [10:00] siretart`: not prohibited, but our tools make it hard to work around DebianMaintainerField [10:00] alkisg: dch -Ui [10:00] Hobbsee: thank you very much [10:00] alkisg: you're welcome. And i'm not sure why that's missing from the manpage [10:01] dpkg-source: Fehler: Version number suggests Ubuntu changes, but Maintainer: does not have Ubuntu address [10:01] dpkg-buildpackage: Fehlschlag: dpkg-source -b hello-2.2 gab Fehler-Exitstatus 255 [10:01] debuild: fatal error at line 1334: [10:01] dholbach: TBH, that sounds like broken tools [10:01] Hobbsee, siretart`: ^ [10:01] well, there you go. And I agree with siretart`. [10:01] dholbach: I don't see that. and I always build ubuntu ffmpeg package on my ubuntu laptop [10:02] siretart`: what is your DEBEMAIL? [10:02] siretart@tauware.de [10:02] Hobbsee: that doesn't solve the problem that bdrung_ has [10:02] siretart`: there you go :) [10:02] ah, you mean it just works because of that [10:02] right [10:27] siretart: done [10:28] <\sh> siretart`: I think I solved the grub problem [10:28] bdrung_: thanks! [10:29] \sh: w00t! :-) - how? [10:29] <\sh> siretart`: when you want to have update-grub working and want to use the ucf technique,...just don't deliver a menu.lst ... do your grub-install in scripts/GRUB/10-setup and $ROOTCMD /usr/sbin/update-grub -y et voila [10:30] yes, that sounds reasonable, if it works [10:30] you rock! :-) [10:30] afk, lunch [10:30] <\sh> siretart`: I can tell you it works right here right now === ApOgEE-- is now known as ApOgEE === ApOgEE is now known as ApOgEE- [11:39] gaspa: when you get bored with checking ocaml give-backs, it looks like octave-* is ready to be checked for give-backs :) [11:39] geser: lol [11:40] I rebuilt a package on an intrepid box, and b/c that box had python 2.5 (standard), it put as a dependency "python (<< 2.6), python (>= 2.5)" (this is the expansion of ' ${python:Depends}' I believe). Now this package won't install on a jaunty box b/c jaunty has Python 2.6 (standard). Any suggestions? (should I manually put more liberal Depends in debian/control for python?) [11:41] up_the_irons: rebuild it on jaunty [11:41] geser: right now i'm after matita ftbfs, then i think i'd move to something else ;) [11:41] NCommander: are you still working on ubuntu on arm these days? [11:41] NCommander: if yes, perhaps you could triage bug #384784? [11:42] Launchpad bug 384784 in mplayer "FTBFS arm" [Undecided,New] https://launchpad.net/bugs/384784 [11:42] geser: was trying to avoid that, it has a lot of build dependencies; but i guess I could as a last resort. === azeem_ is now known as azem === azem is now known as azeem [11:43] up_the_irons: if the package has a compiled python extension you need to rebuild it with python2.6 if you want to use it with python2.6 [11:43] geser: oh ok [11:47] tnx [12:05] What tools are usually used for word-wrapping e.g. the README file in packages? [12:07] personally, i just use vim with ":set tw=72" ;) [12:09] if anyone has time to review and possibly advocate http://revu.ubuntuwire.com/p/pidgin-mbpurple , I will be forever grateful :) [12:10] up_the_irons: hmm that works when writing a new paragraph, but not when editing an existing paragraph (or, *I* can't get it to work...) [12:11] alkisg: gq{ [12:11] :) [12:12] sometimes I have to hit 'J' a lot to bring everything up to the same line, then 'gq{' reformats the whole thing. I'm sure there's probably a keystroke that does it all at once, but i'm too lazy to figure it out [12:12] For now, I'm using Evolution => composing a new text mail => wordwraps, even supports bullets! But it seems kinda silly to keep the original README in the "Draft" email :D [12:12] up_the_irons: ah, that sounds good, let me try... [12:13] alkisg: my vi also supports bullets... ;) [12:13] * alkisg is only basically familiar with vi, must read more... :) [12:14] Thanks, up_the_irons [12:14] alkisg: np [12:15] the more you learn vim, the more everything else just doesn't cut it. unless u like emacs. both are good [12:26] dholbach: Ah, thanks for the pointer about submittodebian. I've just been using reportbug --bts=debian [12:26] it's really good stuff [12:43] is this a jaunty bug? [12:43] libapache2-mod-php5 depends upon apache prefork only === Pici is now known as ZarroBoogs === ZarroBoogs is now known as Pici [12:46] qiyong: What would the bug be? [12:47] libapache2-mod-php5 depends on prefork only, it doesn't work with worker [12:47] php isn't thread-safe, so it's stuck with the prefork MPM. [12:47] that's right. [12:47] It doesn't work with worker. [12:47] if i have to use worker [12:47] what php should I choose? [12:47] Then you can't use mod_php5. [12:48] I use php5 through fastcgi instead. [12:48] Works great. [12:48] ..with the worker MPM. [12:48] libapache2-mod-php5 depends on prefork [12:48] ... [12:48] Yes. [12:49] Because it does. not. work. with. the. worker. mpm. [12:49] so worker is threaded? [12:49] Yes. [12:49] will mod php be threaded one day? [12:49] is threaded the future? [12:49] PHP isself is thread safe, AFAIK. [12:49] The problem is the libraries on which it depends. [12:49] ...which aren't all thread safe. [12:50] So until they're all fixed, you're stuck with the preform mpm (if you *insist* on using mod_php5). [12:50] threaded vs prefork, which is the future? [12:50] Threaded. [12:50] isn't mod_php5 the fastest php? [12:51] With a properly confiured fastcgi setup, you can get comparable speeds. [12:51] * soren realises he's in #ubuntu-motu [12:51] This is totally #ubuntu-server stuff. [12:52] is mod_php4 wokring with worker? [12:52] No. [12:53] And php4 is dead. [12:53] Which depends on the same libraries that php5 does, which still aren't any more thread safe then when php5 uses them. [12:53] then libapache2-mod-passenger is buggy [12:53] it only depends on mpm-worker [12:53] mods work with mpm-worker should be able to work with prefork too [12:55] *shrug* [12:56] I think passenger wants threads, but I'm not sure [12:56] qiyong: i'm running mpm-worker, and installed passenger manually from the gem [12:56] I don't know mod_passenger.. And from the description, I'm going to guess I don't care about it either :) [12:56] qiyong: wait, sorry. i'm running prefork [12:56] I'm happily using it, but I think PHP needs to die [12:57] passenger is quite nice [13:00] up_the_irons: is their other rails mod? [13:01] s/their/there/ [13:01] qiyong: not as an apache module; passenger is pretty much the one to use [13:01] passenger is an apache module, no? [13:01] but there's always the {nginx/lighttpd} -> mongrel approach [13:02] qiyong: yup, apache mod [13:05] which pkg provides php5 fast cgi? [13:07] soren: ^ [13:10] php5-cgi [13:15] with fast cgi, i have to put all my php into cgi-bin? [13:15] soren: ^ [13:16] No, you don't have to change anything, except for a bit of Apache configuration. [13:16] Still... This is #ubuntu-server stuff. [13:16] Please go there. === cemc1 is now known as cemc === savvas_ is now known as savvas [13:43] how do i setup php-cgi? [13:44] qiyong: This isn't a support channel. #ubuntu-server would be the best place to ask. [13:44] or #ubuntu [13:44] and for anything that isn't server related, #ubuntu is the place to ask. Please stop asking non-packaging questions here. [14:18] <\sh> SirDerigo: http://faiwiki.informatik.uni-koeln.de/index.php/FAIUbuntuGrubProblems <- menu.lst + ucf fix ;) [14:19] how do i install a pkg without installing its depends? [14:19] <\sh> SirDerigo: wasn't for you... [14:19] <\sh> siretart: http://faiwiki.informatik.uni-koeln.de/index.php/FAIUbuntuGrubProblems <- menu.lst + ucf fix [14:20] \sh: dont worry [14:23] qiyong, um, but idea, but read man dpkg's section on forcing stuff [14:23] s/but/bad/ [14:24] qiyong: Is there a reason you keep asking support questions in here? You've been asked a few times to take them elsewhere... [14:26] Pici: sorry, some mistakes === yofel_ is now known as yofel [14:47] Heya gang [14:48] heyha bddebian! [14:49] Hi siretart` [14:54] I am packaging a software. Upstream releases a tarball named "appname 1.3.tar.gz" (yes, with a white space). Should I rename it to eliminate the white space or the version number? Or maybe should I just add the orig.tar.gz part? [14:57] lampard: you must not eliminate number. And yes you should rename the tar to remove white space. the .orig.tar.gz should have name in particular format [14:58] lampard: name it 'appname_1.3.orig.tar.gz' [14:59]