[00:00] glad we (and Debian) aren't the only ones with that kind of horrible packaging hack ;) [00:02] god [00:02] now I have to figure out how to do this with cdbs [00:02] I don't suppose you know off the top of your head? [00:03] binary-post-install/foo :: [00:03] sed -i in ... what ion_ said [00:03] sed -ire '...' debian/$(cdbs_curpkg)/foo/bar.pc [00:09] doesn't work [00:09] debian/tmp has been wiped by then [00:10] debian/libdbus-1-dev/... [00:10] it's been dh_install-ed [00:10] it's not referenced by dh_install [00:10] literally the whole of debian/tmp is gone, including the directory [00:10] I hate CDBS [00:10] $ fgrep .pc debian/libdbus-1-dev.install [00:10] debian/tmp/lib/pkgconfig/dbus-1.pc usr/lib/pkgconfig [00:10] yeah I removed that in an attempt to do the sed ;) [00:11] put it back and do 'sed -i blah debian/libdbus-1-dev/usr/lib/pkgconfig/dbus-1.pc' in binary-post-install/libdbus-1-dev [00:11] i.e. let it install and then sed in-place [00:13] sed: can't read debian/libdbus-1-dev/usr/lib/pkgconfig/dbus-1.pc: No such file or directory [00:13] nope [00:13] still fail [00:16] * Keybuk finds a place that works [00:17] I would just put it in install/libdbus-1-dev [00:17] When's binary-post-install? === no0tic is now known as maru0tic [00:18] now I ended up with a .pce file ?! [00:19] o_O [00:20] Oh, -i must be separated from -e etc. [00:20] As in, sed -i.backup -re '...' something [00:20] Oh. Heh.. :) === maru0tic is now known as no0uscia === no0uscia is now known as no0tic [00:25] why bother with -e? [00:25] you don't need it [00:25] nor do you need -r in the example Keybuk gave [00:26] True. I just have -e as a habit. And -r is kind of a default for me. :-) [00:26] I didn’t really take a good look at the actual sed expression. [00:27] it has $ where it should have \$ [00:27] Actually, \$$ since it’s in a Makefile. [00:27] oh yes [00:29] ? [00:31] this looks about right [00:31] binary-post-install/libdbus-1-dev:: [00:31] sed -i 's@-I\$${libdir}@-I$${prefix}/lib@' debian/libdbus-1-dev/usr/lib/pkgconfig/dbus-1.pc [00:31] why use ${prefix} there at all? [00:32] and why \$ ? [00:33] \$ because sed thinks $ means end of line [00:33] ${prefix} because the RH code did ;-) and seems neater [00:37] install/libdbus-1-dev:: [00:37] mkdir -p debian/libdbus-1-dev/usr/lib/pkgconfig [00:37] sed -e 's@-I\$${libdir}@-I$${prefix}/lib@' debian/tmp/lib/pkgconfig/dbus-1.pc > debian/libdbus-1-dev/usr/lib/pkgconfig/dbus-1.pc [00:37] was what I went with in the end [00:40] the version I had was sufficient to get openssh building, at any rate [00:41] if yours produces the same .deb contents (as it looks like it should) then we ought to be fine [00:41] openssh depends on dbus now? :) [00:42] consolekit [00:42] sshkit [00:43] my Bad Idea light is going off [00:43] At this rate, /bin/sh, init and finally the kernel will use dbus. ;-) [00:43] ah, you've screwed it back in after spending a week in the same building as me? :) [00:44] ion_: well, two out of three [00:44] Keybuk: new filament in the bulb [00:45] keybuk: dbus being actually useful for some of them was actually supposed to be part of the joke. :-) [00:45] openssh upstream aren't too happy about the fact that dbus is GPL and openssh is rather aggressively not, which unfortunately I'd completely failed to consider [00:45] so I think I need to figure out how to split that stuff out into a helper [00:45] dbus is attempting to be MIT [00:45] that would help [00:45] it's also AFL [00:46] the AFL is a bit impenetrable [00:46] the MIT thing is held up by one of the companies that contributed having vanished [00:46] apparently this is a major stumbling block, because they can't ok it [00:46] to my mind, if they've vanished hard enough, they can't take legal action either :p [00:47] Heh [00:49] I think (on reading it) that the AFL is MIT-like enough for it not to be a legal problem for Ubuntu; it's literally 10 times the line length though [00:49] (which tends to trip openssh upstream's WHATEVER meter) [00:49] yes, but openssh is not on the rational side of the world [00:50] it belongs in that special camp with other upstreams like glibc [00:53] hmm. on re-reading, describing the AFL as MIT-like is definitely false - it requires source code provision [00:53] (not that I object but it's one thing the MIT/BSD licences determinedly don't do) [00:55] err [00:55] not entirely true [00:55] it only requires source code provision of the original work when distributing it [00:55] it doesn't require source code provision _at_all_ for derivative works [00:56] oh, true [00:56] ugh, though, clause 9 is horrible [00:56] you have to make a "reasonable effort" to obtain explicit assent to the licence [00:56] yeah it's a sucky licence [00:56] which is why they want to change it to MIT/X11 [00:57] is there anything I can quote on that, so openssh upstream don't just get fed up and close the bug? :-) [00:57] on the relicense? [00:58] yeah [00:58] lots and lots of ML posts [00:58] but the conclusion is that it's stuck still [01:00] oh, it's on the website, I can quote that [01:00] http://lists.freedesktop.org/archives/dbus/2008-February/009410.html [01:01] the current status is that for almost all of the code, the authors have agreed to an MIT/X11 licence [01:01] (including us! :p) [01:02] ta [01:13] random use of bzr #15 [01:13] import two version of a package [01:13] and use bzr blame to figure out which lines haven't been changed [01:18] like inverse diff? [01:18] yeah [01:18] patchutils doesn't have one ;) [01:19] /usr/bin/same :p [01:19] Heh [01:22] hello. I've submitted a patch upstream (linux-kernel) for detection of macbook pro 4,1 and macbook air keyboards. It'd be nice if it made it into ubuntu sooner. Should I resubmit it for ubuntu, and if so, where? [01:27] ubuntu-kernel [01:30] is that a mailing list? what's the full address? Is there a page that details the patch format for ubuntu etc? [01:32] mailing list [01:32] same general method as lkml [01:33] hm https://lists.ubuntu.com/mailman/listinfo/ubuntu-kernel says no such list [01:33] grep for kernel ;) [01:34] kernel-bugs? [01:34] kernel-team [01:34] ok. thanks guys [01:35] new to this stuff... heh [01:35] https://wiki.ubuntu.com/KernelGitGuide [01:37] can I just resend the patch I sent upstream? [01:44] probably [01:48] done. thanks again :) === asac_ is now known as asac === asac_ is now known as asac [03:14] What does it mean when kill -9 won't kill an process? :P [03:15] It's probably in D state? It'll be waiting in the kernel, and can't recieve signals at all. [03:16] doh, it is in D state [03:16] Stupid firefox :/ [03:17] Yup. Unkillable. [03:17] Weird... new applications won't start :/ [03:17] D state makes the baby jesus cry. [03:18] Now all my applications are going into D state :/ [03:18] Right. Something has died horribly, and the world is ending. [03:20] Reboot is only option? [03:21] If you can get /sbin/shutdown to not D state, yeah :) [03:22] slangasek: thanks for merging ;-) [03:22] n/p [03:23] nooo!!! :( [03:23] * cody-somerville just lost the Xubuntu Strategy document draft 2. === asac_ is now known as asac === pabluk_ is now known as pabluk [05:52] Good Afternoon. [05:54] I was sitting briefly, aghast 'n pondrance, and it became notable to one that there is no cron gui management tool, that one of course being me; if you could so explain some burdened parly to me in the form of a hairdresser in silk stockings, I'd be very delighted, or possibly intrigued to find a solution of some sort. [05:56] ? [06:04] DrVague: that's probably because any gui-driven job scheduler does far more than cron, and the gui comes after the decision that cron is lacking [06:04] lamont: what? you're suggest a replacement backend for cron? [06:05] you're kidding? [06:05] no. I'm saying that a gui for cron is the least of its issues [06:05] if you expose cron to a gui-requiring user, then you also have to teach him shell scripting... [06:07] hurrrrr? [06:07] that doesnt follow for me [06:07] well, what shell commands is he going to put in his crontab? [06:08] i was thinking he would just use the gui to schedule actual sh scripts to run and set the time with a mouse [06:08] 'add new job' ::file selection dialog:: 'time' ::calendar:: [06:08] and then the first thing that anyone who's used an actual job scheduler on (say) a mainframe runs into is "how do I tell it to release job B when job A finishes??" (answer? run a real job scheduler instead of cron, or hack together your own scheduler in the jobs) [06:08] 'save' 'cancel' [06:09] right. and someone has to write those shell scripts... [06:09] and whoever writes those generally drops something to run them into /etc/cron.* [06:09] thats like saying the team who maintains geany should be responsible for teaching everyone perl [06:10] my point is more that anyone who is satisfied with cron in the current world doesn't use a gui, and there is zero motivation to provide one for the developers [06:11] tbh, cron is used for the same, probably 2 dozen unique job scripts anywhere, a pulldown menu would cover most peoples' needs [06:11] rather, no motivation for the developers to produce one, since the user community divides neatly into the group that doesn't see the need, and the group that would ask lots of questions of whoever wrote the gui. [06:11] oh, yeah, well that is def true [06:12] which tends to discourage people from bothering with that, when there are far more interesting/significant things to work on that don't immediately become a support nightmare. [06:12] hrm. i hate that community gap between devs and recreational users. [06:12] good morning [06:12] morning dholbach [06:12] morn' [06:12] morning, lamont, Daniel [06:13] hrm... I think that means it's about bedtime. :-) [06:13] hi lamont DrVague Mithrandir :) [06:14] lamont: hehe. [06:14] well, given that the workday starts in just under 7 hours... [06:15] hrm.. importing a 300MB, 5900 file cvs repo into a real VCS? not so fast. [06:15] are you surprised? [06:16] nope [06:16] I just hope it finishes [06:32] Good morning [06:34] Morning pitti. [06:35] pitti! [06:35] hi pitti, TheMuso, Hobbsee [06:35] Heya dholbach. [06:38] hello pitti [06:51] Hi pitti, Hobbsee, TheMuso, ajmitch [06:52] Anybody got any idea why /dev/mapper contains all of my LVM volumes, but /dev/ doesn't? [06:58] wgrant: /dev/vgname isn't even supposed to exist, AFAIK [06:59] pitti: sbuild seems to want it. [06:59] Well, schroot. [07:06] don't schroot, I'm unarmed [07:16] Riddell: done now [07:42] hmmm [07:43] if you install slocate and run update-alternatives to select it as the preferred locate (over mlocate) [07:43] then the slocate cron.daily script still doesn't fire because it has the following guard [07:43] if [ -x /usr/bin/slocate ] && [ ! -x /usr/bin/mlocate ] [07:43] sdh: that was done to avoid running slocate on gutsy->hardy (or dapper->hardy) upgrades [07:44] since we don't want/can't automatically uninstall slocate completely [07:44] pitti: interesting... why not consult alternatives and act based on that? [07:44] i did a gutsy -> hardy upgrade, and got mails from cron like this: [07:44] /usr/bin/updatedb.mlocate:/etc/updatedb.conf:4: unknown variable `FINDOPTIONS' [07:45] /usr/bin/updatedb.mlocate:/etc/updatedb.conf:5: unknown variable `export' [07:45] etc... which i assume is mlocate trying to parse slocate configuration [07:45] noone went through that much trouble, I guess, since slocate is basically deprecated [07:45] right, that would be it [07:45] hmm [07:45] sdh: can you please file a bug about it? it's worth it, at least [07:45] pitti: i still have to work out exactly what happened ;-) [07:45] but yeah, sure [07:46] now i'm confused: [07:46] % dpkg -S /etc/updatedb.conf [07:46] mlocate: /etc/updatedb.conf [07:48] i won't file a bug on that until i can understand what is going on [07:49] but i note that somebody said "8.04 uses mlocate instead of slocate on desktops now" -- but it seems it is used on servers too, which imho negates the reasoning behind deprecating slocate in favour of mlocate [07:49] (i.e. affecting user experience especially on laptops) [07:49] sdh: both slocate and mlocate ship this configuration file [07:50] on hindsight, we should have probably made the two conflict to each other [07:50] sounds like a plan [07:50] (we can still do that in -updates, at least) [07:51] cool [07:51] sdh: may I ask why you prefer sloacte? [07:51] slocate, even [07:52] Mithrandir: it's not that i prefer slocate - though i do think, in my case "if it's not broken, don't fix it" - it's that i dislike getting error messages from crond when the locate implementation is changed [07:52] sdh: oh, that I can agree with. [07:52] i see that i've been using mlocate unknowingly on my hardy laptops for the last few months anyway and i have no problem with it :) [07:54] yeah, I can't see anything slocate gives us over mlocate so I was wondering why anybody would want to continue using slocate. [07:54] sentimental reasons ;-) [07:55] oh god the whole mlocate/slocate thing is a mess [07:55] i have /etc/cron.daily/{mlocate,slocate,find} [07:56] i think that's https://bugs.launchpad.net/ubuntu/+source/mlocate/+bug/197170 [07:56] Launchpad bug 197170 in mlocate "cron daily runs updatedb twice" [Undecided,Confirmed] [07:58] * sdh wonders why findutils has cron.daily/find doing an updatedb [07:58] sdh, so that the locate command is useful; there's been discussions of getting rid of that [08:00] liw: i thought locate (1) was just an alternatives link into mlocate/slocate [08:00] man i am confused now, i have mlocate installed and seem to have no locate (1) [08:00] sdh, locate is part of find, and the cron job runs regardless of what the alternatives say (which, as it happens, could be considered a bug: please file one :) [08:01] liw: trying to get my box working first :) any chance you could dpkg -S bin/locate for me please? i appear to have lost mine [08:03] findutils, apparently. odd that i don't have it, then. [08:03] I don't seem to have it either [08:03] hmm [08:03] this seems like a bit of a mess [08:03] yes! b0rkage! [08:03] did you do any mlocate/slocate manual tweaking? [08:03] no [08:04] steve@ace:~% dpkg -l '*locate*' [08:04] No packages found matching *locate*. [08:04] steve@ace:~% dpkg -S bin/locate [08:04] findutils: /usr/bin/locate [08:04] actually ignore that, that's a gutsy box! :) [08:04] but the problems i've been discussing are on a hardy (from gutsy) box [08:05] right, having reinstalled findutils it definitely contains no locate(1) [08:05] > dpkg -S bin/locate [08:05] locate: /usr/bin/locate.findutils [08:05] on hardy [08:05] *boggle* [08:05] /usr/bin/locate is handled as an alternative [08:06] ah yes [08:06] % update-alternatives --list locate [08:06] /usr/bin/mlocate [08:06] % locate [08:06] zsh: command not found: locate [08:07] sorry for the pasting btw, is it excessive yet? [08:07] nah, no problem when it's just a line or two at a time. [08:07] ;> [08:08] hm, I think this machine was installed with hardy, not gutsy. [08:08] % dpkg -L mlocate | grep 'bin.*locate' [08:08] /usr/bin/mlocate [08:08] /usr/bin/updatedb.mlocate [08:08] % sudo update-alternatives --config locate [08:08] There is only 1 program which provides locate [08:08] (/usr/bin/mlocate). Nothing to configure. [08:08] % locate foo [08:08] zsh: command not found: locate [08:08] you might need to rehash [08:08] that path is wrong [08:08] steve@nemesis:~% rehash [08:08] steve@nemesis:~% locate foo [08:08] zsh: command not found: locate [08:08] ;) [08:08] it's the path in the alternatives [08:09] what does ls -l /etc/alternatives/locate say? [08:09] or, hang on [08:09] lrwxrwxrwx 1 root root 16 2008-04-03 15:21 /etc/alternatives/locate -> /usr/bin/mlocate* [08:09] is what I have [08:09] erk! [08:09] % ls -l /etc/alternatives/*locate* [08:09] lrwxrwxrwx 1 root root 16 2008-05-30 07:42 /etc/alternatives/locate -> /usr/bin/slocate [08:10] try update-alternatives --auto locate ? [08:10] that seems to have fixed it [08:10] but why is u-a telling me it points to mlocate when it points to slocate, i wonder? [08:10] it's not, it's just telling you there's nothing to configure [08:11] good point > [08:11] I wonder if this is really an update-alternatives bug. [08:11] i am inclined to agree [08:11] can you try installing slocate, changing the alternative to point to slocate, verify it's in manual mode pointing to slocate, then purge slocate? [08:11] it shouldn't say "only 1 program, nothing to configure" if it is in fact handling it, badly :) [08:11] and then see what it points to? [08:12] though i'm reluctant to keep screwing with my production boxes... sure! :) [08:12] Mithrandir: step 2 using u-a --config locate, right? [08:12] yes [08:12] no manual hacking in /etc/alternatives. ;-) [08:13] Using '/usr/bin/slocate' to provide 'locate'. [08:13] lrwxrwxrwx 1 root root 16 2008-05-30 08:12 /etc/alternatives/locate -> /usr/bin/slocate [08:13] now to purge [08:13] Mithrandir: Why not? [08:13] % sudo update-alternatives --list locate [08:13] /usr/bin/mlocate [08:13] lrwxrwxrwx 1 root root 16 2008-05-30 08:13 /etc/alternatives/locate -> /usr/bin/mlocate [08:14] seems good now [08:14] soren: in this case, because I want to see if u-a DTRT. [08:14] Mithrandir: Oh :) [08:14] sdh: hmm, ok. [08:14] Mithrandir: strange though... [08:14] Mithrandir: would you expect an apt-get remove, without --purge, to update alternatives? [08:14] Mithrandir: let me try that again without purge. [08:14] sdh: sure, try. I think it should. [08:15] since you should never be left with a dangling alternatives symlink. [08:15] but you said you were left without /usr/bin/locate at all, or was it just dangling? [08:15] hmm that seemed to work too, odd. [08:16] Mithrandir: hard to tell, now that i've messed about! it gave me no such f/d, but that could be either [08:16] looks to me like it's just a gutsy/hardy upgrade quirk [08:17] but it does feel like a can of worms, including *locate and even u-a [08:17] you don't happen to have a backup of the system you could take a look at a directory listing from? [08:17] and i definitely think that mlocate/slocate should be mutex [08:17] yeah, it looks like it might be ordering dependent. [08:17] Mithrandir: not in any kind of realistic timescale, afraid not [08:17] which is.. ugh. [08:17] ok. [08:19] thanks for helping out [08:19] oh, happy to help; I still wonder how we can get it properly fixed though. [08:20] yeah but it's at that awkward stage where i'm not sure how to reproduce the problem [08:20] it might be related to /usr/bin/locate being a proper file in gutsy. [08:20] i'm not even 100% sure what the problem was [08:21] i have to get to work soon... but i suppose i have time to break out a few VMs :) [08:21] if it's present on a stock install + upgrade, it should be fairly easy to see what's happening. [08:22] i conveniently have a gutsy template vm [08:31] right Mithrandir... i have a gutsy box ready for upgrade to hardy [08:32] i'll install slocate then do the upgrade [08:32] sound like a reasonable test to you? [08:33] sounds good [08:33] just to confirm, on gutsy /usr/bin/locate is a symlink to /usr/bin/slocate [08:33] (with slocate installed, obviously) [08:34] *nod* [08:34] before that it was provided by findutils [08:40] it's chugging away :) [08:41] how would i go about messing with [remapping] keys [08:53] Mithrandir: well it came back with mlocate installed and slocate uninstalled, which i guess is right [08:53] and the alternatives link points to mlocate [08:53] ...so it all seems ok. [08:57] hmm [08:57] so it's.. something else or harder. [08:58] seb128: WDYT about bug 182945? [08:58] Launchpad bug 182945 in gio-standalone "gio-standalone should be remove, gio is in glib now" [Undecided,Confirmed] https://launchpad.net/bugs/182945 [08:58] seb128: no reverse dependencies [09:01] pitti: it should have been removed before hardy, that was a temporary package for gio before having it in glib [09:01] seb128: merci; removing then [09:01] thanks [09:02] pitti: I just uploaded a new rhythmbox revision, I attached the debdiff to one of the zillion bugs it closes, is that good enough or should it be attached to all of those? [09:02] seb128: one is enough [09:03] what's the best way to learn how to program? i've taken a basic college course in C & C++, but I wanna look at real programs... even if they're mildly confusing at first [09:03] any suggestions? [09:11] what program has easy to read source code for a n00b?? [09:12] dolphin: there's a good book... let me find it [09:12] http://www.spinellis.gr/codereading/ [09:16] sdh: aww, man... you gotta pay for the whole thing?!? =-( [09:17] dolphin: it's worth it [09:18] dolphin, pick a program in Ubuntu that interests you, then do "apt-get source foo" to get the source, and have at it; it's probably best to pick a simple command line utility to start with, they tend to be simpler [09:18] fileutils or something is interesting [09:18] and easy to see how code relates to runtime [09:18] yea, i was thinking graphical... but text is definitely a better start [09:19] dolphin: if you want graphical then you have to get familiar with the intricacies of whatever graphical toolkit is being used, be it gtk/qt/etc [09:19] dolphin, the coreutils package might be a start, for example: cat, rm, cp, and such, although they have a lot of complexity by having to deal with all sorts of quirks in various Unix flavors [09:19] dolphin: that is hardy, so i'd start with text [09:19] err, s/hardy/harder/ [09:19] findutils is a good choice [09:20] $51 isn't that bad, i guess i'll order that book to keep me from getting too bored this summer [09:20] dolphin: it seems there is a sequel but i haven't read that [09:21] what is fileutils? [09:21] dolphin: but code reading is a good book because it talks you through various idioms in real software (e.g. apache, netbsd) that means you can make the leap from understanding, say, C syntax, to being able to understand large chunks of code [09:21] dolphin: sorry, i meant coreutils i guess [09:21] ahh... k [09:21] the package with head/tail/wc etc [09:22] this is definitely the starting point i'm looking for... i'm tired of simple labs [09:22] right [09:22] so between apt-get source and that book, i'd say you could spend a lot of useful time understanding and then modifying code [09:22] thanks for the recommendations, ya'll! [09:22] np [09:23] gonna catch some ZZZzz's for now.... lata [09:23] thanks again! [09:23] the nice part of going to apt-get source way is that you can easily modify code too, and try out your modifications for real [09:23] (running, say, your modified corutils in a virtual machine may give additional confidence) === thekorn__ is now known as thekorn [09:38] pitti: there's a packagekit discussion going on about offering to install relevant drivers when a piece of hardware is plugged in, does jockey take care of that for us? [09:39] I mean does it get triggered by hal, or does it just scan at startup? [09:39] james_w: not quite yet, but getting there [09:39] james_w: ATM it just scans at session start [09:39] so it's planned, and I can ignore trying to implement it for packagekit? [09:39] james_w: but for 0.5 I plan to hook it into hal [09:39] thanks [09:39] james_w: yes, pretty much; we'll do it the other way round, we'll use PK to install the drivers from jockey :) [09:40] heh, works for me :-) [09:40] james_w: and jockey will go into Fedora soon, too [09:40] so I'm not even sure whether they should do that in PK itself (one tool for one purpose, etc.) [09:40] but *shrug* [09:40] oh, cool. Apport as well, they're loving your code. [09:52] pitti: james_w: atm, gnome packagekit has a GpkFirmware GObject that watches udev for missing firmware requests and prompts to install it if finds then in a source [09:52] sorry to keep bothering you [09:52] james_w: (no problem) [09:52] james_w: that sounds nice; however, our packages don't support that ATM [09:52] james_w: most firmware should already be present in l-r-m [09:53] and if someone uninstalls l-r-m, he won't want the non-free firmware IMHO [09:53]