/srv/irclogs.ubuntu.com/2012/01/09/#ubuntu-devel.txt

=== jasox is now known as jasox_afk
=== jasox_afk is now known as jasox
pittiGood morning08:34
ajmitchmorning pitti08:35
sladenmorning08:36
cyphermoxgood morning08:37
evhiya08:49
gesergood morning08:51
pittiev, cjwatson: I've got working GI bindings for libxklavier yesterday; I submitted them upstream, but if you want to work on it soonish, I can create a PPA for you if you want?08:55
pittiev, cjwatson: that's for bug08:55
pittibug 80056108:55
ubottuLaunchpad bug 800561 in libxklavier (Ubuntu Precise) "No way to add other keymap than english on Live CD" [High,In progress] https://launchpad.net/bugs/80056108:55
evpitti: cheers! I'm happy to wait for it to land in the archive, given that we shipped 11.10 without it08:56
pittiev: upstream already responded yesterday, I hope it'll land RSN; but in case you want to play around with it, I wanted to tell you08:58
evthanks09:10
mptmvo, hi, is robert_ancell around? I'd like to talk with you when you're both free, about recovering from failed upgrades09:35
mvompt: I check the desktop room09:44
pittimpt: in the desktop room here09:45
mptHi robert_ancell, are you or mvo on Skype? Or else Mumble?09:49
robert_ancellmpt, we can mumble09:49
robert_ancellmpt, mvo is busy at the moment, do you want both of us?09:52
mptrobert_ancell, ideally.09:52
robert_ancellok, we'll call then09:52
mptthanks09:52
robert_ancell(in 15 mins)09:52
mptok09:53
cjwatsonpitti: that's great, thank you - what ev said09:57
mvompt, robert_ancell: I'm ready now, sorry for the delay10:26
mptmvo, robert_ancell was just on Mumble but then left10:27
mptrobert_ancell, phone call instead?10:32
mptmvo, sorry, I'm not hearing what you're saying10:36
mvompt: not hearing at all or in bad quality?10:37
robert_ancellmvo, now your mic is stuck on10:37
mptmvo, you sound like a MIDI keyboard10:37
mvompt: I think we need both, at startup and at the installer level10:37
mptmvo, ok10:37
mvompt: as the startup maybe in susch a bad shape that it can not recover from the installed system10:37
mptmvo, robert_ancell: So would the UI for the startup variant be inside Plymouth?10:38
robert_ancellmpt, I would have thought it would be in grub or failsafe-x10:38
mvowe have support in friendly-recovery for recovering from a broken upgrade, but the UI is certainly lacking :/10:39
mptIs it a console UI?10:39
mvoyes10:39
mvotext ui10:39
mvonot even progress bars10:39
mptok10:40
mptIf they were both implemented, would the startup and installer-based recovery systems share code at all?10:41
mvothat is a interessting question, it depends on how we do it - but I would image that the installer-based on is very much apt-clone based. I haven't put much thought into how it would be done on startup, but there are certainly area where code can not be shared, like the UI in ubiquity would be different etc10:43
mptmvo, that question was mainly about whether it should be one or two specifications :-)10:45
mptmvo, robert_ancell: so, I'll do sketches for both a Ubiquity-based UI and a console-based UI10:47
robert_ancellmpt, ok, so mvo and I talked here and we think you probably just need an ubiquity UI and a software-center UI10:53
mptrobert_ancell, why would it go in USC?10:53
mvompt: and the software-center UI we have (the rapair broken catalouge)10:53
robert_ancellmpt, so ubiquity is used for severe corruption cases, and software center for minor breakages (e.g. flash)10:54
robert_ancellThe stuff we covered at UDS (https://wiki.ubuntu.com/GracefulFailure) is more about detecting failure and notifying the user as opposed to necessarily fixing the problems10:54
mvothe console based one would fall right into the middle of this so its probably not worth the effort10:54
mptok10:55
robert_ancelland it's not clear which part of the boot would be able to detect the failure (without side effects of slowing boot) and launch such a console UI10:55
mvofor the console UI you would have to have a work kernel, python at least10:58
seb128mvo, bug #91371911:11
ubottuLaunchpad bug 913719 in update-manager (Ubuntu) "gets confused by packages in the dpkg "in" state" [Low,New] https://launchpad.net/bugs/91371911:11
mvothanks seb12811:13
seb128mvo, thank *you* ;-)11:13
mptmvo, how long would it take to go through the available partitions on a typical PC, and detect that one of them was a broken Ubuntu installation? Seconds? Minutes?11:28
mvompt: ev will be able to answer this better, but I think second11:28
mvos11:28
mptok11:28
evanswered out of band11:31
=== _salem is now known as salem_
pitticjwatson: are you aware of any recent changes in the live system build? desktop CDs are now 21 Mb oversized, and I think last week they were at 705 MB; but the alternates did not grow, so iso-deb-size-compare won't work11:54
pitticjwatson: hm, then again the health check from an hour ago still says11:56
pittiubuntu/daily-live: precise-desktop-amd64.iso oversized by 5550080 bytes (739553280)11:56
pittiso maybe I mix it up with that (wrong?) number11:57
cjwatsonpitti: not that I know of12:18
Riddellpitti, cjwatson: is there a tech board meeting today?12:38
cjwatsonRiddell: good question, I guess it depends on what the schedule in Budapest is like12:54
pittiI guess I can make it12:55
pittiwho is chairing today?12:56
pittiah, mdz12:56
=== jdstrand1 is now known as jdstrand
=== l3on__ is now known as l3on
MacSlowpitti, ping13:51
pittiMacSlow: hey13:51
MacSlowpitti, hey... I just switched to precise on my laptop and not compiling libunity it complains about missing "/usr/lib/i386-linux-gnu/libgio-2.0.la"13:52
pittiMacSlow: right, .la files are gone; we don't need them, they only cause unnecessary extra dependencies13:53
MacSlowpitti, it's form some part of the linker-work during make... I was wondering if there might be an issue with the packages on precise13:53
MacSlowpitti, so what's needed on my side now to make the "make" work again?13:53
pittiMacSlow: it might come through a library in between, though; perhaps you can pastebin the last couple of lines of build log?13:53
pittii. e. a faulty .pc file of a library package13:54
MacSlowhttp://pastebin.ubuntu.com/798218/13:55
pittiMacSlow: seb128 is coming over to you13:55
pittiMacSlow: most probably you have a local broken /usr/local/lib/*.la13:55
MacSlowpitti, cool... seb128 thx :)13:55
=== bladernr_afk is now known as bladernr_
barryScottK: re: "port to qt" - not sure either, but mvo mentioned it when we chatted.  that was mostly just a placeholder.  if it's available upstream, pending packaging, that's much better14:31
ScottKCool.14:31
barrythanks for following up on the blueprint14:32
ScottKI played with it a little and it's not 100% straightforward to package, but I think it's doable with a bit of fiddling.14:32
barrydo you have a branch or other code i can start with?14:32
=== bladernr_ is now known as bladernr_afk
ScottKNo.14:32
barryk14:33
ScottKI'd start with pykde4 in precise.14:33
barryyep14:33
ScottKbarry: The one thing that's not just coercing it to build is figuring out what to do about /usr/lib/kde4/kpythonpluginfactory.so.14:34
ScottKAlthough with the .so name munging in python3, maybe it's OK.14:35
=== bladernr_afk is now known as bladernr_
=== bladernr_ is now known as bladernr_afk
=== bladernr_afk is now known as bladernr_
barryScottK: right, it would end up being kpythonpluginfactory.cpython-32mu.so so it shouldn't conflict *although* the import machinery might get confused if they both exist15:04
ScottKThat would be the tricky part.15:05
ScottKWill python3.2 prefer a munged name over an unmunged one?15:05
ScottKIf it will, it should be ~fine.15:05
barryScottK: i'm checking the source now :)15:06
ScottKCool.15:06
barryScottK: yep, module.soabi.so gets found before module.so, so we should be good (glad i did the right thing and/or used the time machine keys)15:15
ScottK:-)15:15
ScottKThen it's probably just a matter of build-deps and beating cmake into submission.15:15
l3ondoko, I looked at libkml, but your patch introduced in bug debian #650525 does not work15:16
ubottuDebian bug 650525 in libkml "ld-as-needed patch needs an update" [Normal,Open] http://bugs.debian.org/65052515:16
l3oncan you help me to understand wath's wrong? :)15:16
l3on*what15:16
barrycool, now it's in the whiteboard so i'm on the hook :)15:17
ScottKExcellent.15:17
tjaaltonslangasek: got a multiarch-related bug for you, not sure where to put it, bug 87414315:17
ubottuLaunchpad bug 874143 in libpciaccess (Ubuntu) "package libpciaccess0 0.12.1-2 failed to install/upgrade: ErrorMessage: libpciaccess0:i386 0.12.1-2 (Multi-Arch" [Undecided,New] https://launchpad.net/bugs/87414315:17
tjaaltonslangasek: so apparently the non-native version of the package got installed before the native one got upgraded15:20
tjaaltonsee the end of apttermlog15:20
dokol3on, to reproduce it yourself, you could try to build in a oneiric or precise chroot (debootstrap sets it up for you). do you package a new upstream version?15:27
l3ondoko, no. I have grabbed merge from debian15:28
dokol3on, build log?15:29
l3onlost, wait a bit, I'm going to rebuild it15:30
trismIs there a reason the debug dbus-daemon binary is installed to /bin? (in both oneiric and precise)15:40
l3ondoko, http://debomatic64.debian.net/precise/pool/libkml_1.3.0~r863-3ubuntu1/libkml_1.3.0~r863-3ubuntu1.buildlog15:43
l3onthe current debian package has your patch on board15:43
dokol3on,  ../../src/kml/engine/libkmlengine.la ../../src/kml/regionator/libkmlregionator.la15:45
dokoexchange these two15:45
slangasektjaalton: this is bug #83562515:48
ubottuLaunchpad bug 835625 in apt (Ubuntu Oneiric) "apt may try to unpack a foreign-arch multiarch library before the native package is at a multiarch version, prohibited by dpkg" [High,Fix released] https://launchpad.net/bugs/83562515:48
tjaaltonslangasek: cool, thanks, I'll dupe it then15:49
slangasektjaalton: I don't think we're going to backport to 11.04 apt though, because we weren't provisioning multiarch for users in 11.0415:49
slangasektjaalton: so you've been hit by this problem by being an early adopter only15:50
tjaaltonslangasek: nah it wasn't me, just was going through the bugs against libpciaccess15:50
slangasektjaalton: ok15:51
semiosisWhen/how often does Ubuntu sync from debian unstable?15:55
cjwatsonThis cycle we've been syncing from testing, not unstable15:56
cjwatsonWe just stopped doing so today15:56
cjwatsonFor the first half or so of a release cycle, the sync is roughly daily15:56
semiosiscjwatson: when you say stopped doing so... does that mean you've stopped debian syncs altogether or have switched to unstable?15:58
semiosishere's my issue, i work with the debian maintainer of the glusterfs project and we just committed a patch that enables the latest version to build.  i'd like to see that latest version included in 12.04.  should I now file a syncrequest?15:59
ScottKYes.15:59
cjwatsonsemiosis: We stop syncs for the second half of every Ubuntu release cycle15:59
cjwatsonBut just the automatic ones - you can still request them15:59
semiosisah ok.  thanks i'll do that.15:59
cjwatsonSo yes, file a sync request if you want it15:59
semiosisone more thing... the debian package glusterfs-server includes an initscript, but that causes a bug on ubuntu beucase of upstart/mountall.  i've contriubted an upstart job to the upstream project, but since debian doesn't use upstart, its not in the debian package.16:00
semiosisso i'd like to add a merge to ubuntu to add the upstart job to the glusterfs package in ubuntu16:00
semiosisdoes that sound right?16:01
ScottKYes.16:01
ScottKAlthoguh upstart is in Debian, so there's no reason it couldn't be added there too.16:02
cjwatsonupstart jobs are a fairly common source of divergence at the moment though, so *shrug*16:02
semiosisafaik including an upstart job in the debian package adds a requirement on upstart-job, which causes install to break on debian boxes16:02
cjwatsonyou wouldn't be the first16:02
cjwatsondepends on how you do it ... but yeah16:02
trismWhen I say debug dbus-daemon, I don't mean unstripped, I mean, has the DBUS_BUILD_TESTS sections enabled. It is copied twice in the build logs, once from build/bus and then overwritten by build-debug/bus16:03
=== yogi is now known as signbit
l3on__doko, → http://debomatic64.debian.net/precise/pool/libkml_1.3.0~r863-3ubuntu1/libkml_1.3.0~r863-3ubuntu1.buildlog16:15
l3on__cp: cannot stat `debian/tmp/usr/share/java': No such file or directory16:15
semiosiscjwatson, ScottK: thanks again!16:16
=== l3on__ is now known as l3on
stokachuslangasek: I was looking into bug #858122 comment #45 and didn't see anything related in the bazaar revisions that would relate to this. I see wfink has been doing most of the commits would he be able to provide some status on that bug?16:20
ubottuLaunchpad bug 858122 in sysvinit (Ubuntu Precise) "incomplete migration to /run (shutdown script order has been demolished)" [High,Triaged] https://launchpad.net/bugs/85812216:20
stokachuslangasek: if nothing has been done I dont mind picking it up and looking into i16:20
stokachuit*16:20
slangasekstokachu: wfink is upstream for insserv, he wouldn't have any input on this as it's entirely a distro integration question16:22
stokachuah16:22
stokachui may be looking at wrong package in launchpad then16:23
slangasekstokachu: the main thing to do be done is to move insserv off the system path, I believe16:23
slangasekstokachu: I think you're looking at an import of the upstream source rather than at the package16:23
stokachuok, so we are thinking move it off system path then help pkg maintainers to fix their packages?16:23
stokachuor just provide a technical release note of some sort16:24
slangasekwhich package maintainers do you mean?16:24
stokachuthese would be 3rd party like turboprint who calls insserv directly16:24
slangasekdoes turboprint call insserv unconditionally, or only if detected?16:24
stokachugood question I'd need to get their source packages and look into it16:25
slangasekFWIW I wasn't planning on doing any work to fix third-party packages; they're using a non-standard interface, and stopping that interface from being available fixes the critical bug16:25
stokachuok sounds good16:26
stokachuslangasek: would I be wrong in suggesting upstart replace/conflict: chkconfig, insserv?16:40
stokachuupstart already replaces sysvinit so why not go a bit further16:40
slangasekstokachu: I don't know that the chkconfig package is incompatible, is it?  And as for insserv, upstart itself doesn't conflict with insserv at all, and in the long term I actually think we will be using insserv for bridging upstart and sysvinit dependency information16:48
slangasekstokachu: so I'd really prefer that we just shunt insserv to a different directory and not diverge from Debian's initscripts package for this16:48
stokachuok16:48
stokachuah yea so update.rc-d uses insserv... i think chkconfig calls insserv directly though, i need to research that more16:49
cjwatsonmvo: so for https://blueprints.launchpad.net/ubuntu/+spec/foundations-p-lts-upgrades, the way I did this for previous releases was to copy .debs into dists/$CODENAME/main/dist-upgrader/binary-$ARCH/ on the CD - will that still work?16:56
mvocjwatson: it should, yes16:56
cjwatsoncool, that should be easy then16:58
mvocjwatson: let me know once its there and I will a) re-enable my old code b) test ;)17:05
=== jasox is now known as jasox_study
=== bladernr_ is now known as bladernr_afk
=== bladernr_afk is now known as bladernr_
Riddellmdz, pitti, cjwatson: well I need to go out, if there is a tech board meeting I can get home for it fine but please text message me 20 mins before17:17
Riddelljriddell.org/contact.html17:18
mdzRiddell, I'm expecting to have a meeting; is there talk of canceling it?17:30
mdzthere is content on the agenda17:30
ScottKmdz: It looks like last meeting's content though. I think PAE kernel is done.17:32
ScottKPAE/non-PAE17:34
Riddellmdz: there's no agenda made and pitti and cjwatson weren't sure if it was happening17:34
mdzScottK, Riddell, two items were just added today17:44
Riddellok I'll assume Kubuntu LTS is being discussed and be back in time for it17:45
=== Quintasan_ is now known as Quintasan
argesbroder, hello. does your backportpackage script allow for first backporting a dependency and then feeding that to backport the original package? or is this something I need to set up in pbuilder? thanks17:47
=== jasox_study is now known as jasox
=== bladernr_ is now known as bladernr_afk
=== andreas__ is now known as ahasenack
psusicjwatson: since partman makes an extended partition for you anyhow, why doesn't it just put the root filesystem inside it as well, instead of only the swap partition?19:43
cjwatsonpsusi: Because many BIOSes will refuse to boot if you have no primary partitions at all.19:45
funkyHatre. the new version of gnutls (https://lists.ubuntu.com/archives/ubuntu-devel/2012-January/034629.html)... surely LGPLv3+ allows use in GPL2 projects? Surely that's the point of the LGPL (well not specifically GPL, but other licenses, including closed source ones)19:45
psusicjwatson: ahh, so makes sense if you chose the option to use the whole drive, but when installing along side windows...19:47
cjwatsonpsusi: Then it will use logical partitions for both.19:47
psusicjwatson: hrm... from what I've seen it still puts the root in a primary and only swap in logical19:48
cjwatsonpsusi: Happy to look at a partman log, but it's not supposed to do that in general.19:49
cjwatsonfunkyHat: I'm afraid that's not the way it works19:49
psusicjwatson: good to know... will file a bug report with the logs if I see it again19:49
cjwatsonfunkyHat: you can google for "lgplv3 gplv2 compatibility" for instance ...19:50
cjwatsonfunkyHat: http://gplv3.fsf.org/dd3-faq has a compatibility matrix19:51
funkyHatcjwatson: thanks. I'll have to re-read the licenses to try to understand why...19:53
cjwatsonfunkyHat: if it helps, the reason for the incompatibility is in the terms of the GPLv2, not in the terms of the LGPLv320:03
funkyHatcjwatson: yeah I found a couple of bits relating to samba that said that. I didn't realise GPLv2 was that restrictive20:04
cjwatsonfunkyHat: GPLv2 says "must distribute under terms of GPLv2 with no additional restrictions" and LGPLv3 has a couple of additional restrictions over GPLv220:04
cjwatsonIIRC20:04
cjwatsonthere's the system libraries exception, but we don't get to use that when we're distributing both library and thing-that-uses-library20:05
funkyHatSo GPLv2 software can't link with libraries under *any* license that is more restrictive than the GPLv2 (with the exception of standard system libraries, apparently)20:05
cjwatsonright (and even then only if distributed separately from those system libraries)20:07
funkyHatI suppose because I've seen GPL software released for Windows I assumed it was ok to link with anything, where in fact it's only ok because of the system libraries exception20:07
cjwatsonthis is also the reason the GPL is incompatible with the OpenSSL licence20:07
cjwatsonfunkyHat: yes20:07
cjwatsonmy understanding is that Microsoft would not be permitted to distribute GPLv2 software together with Windows, although I haven't previously thought about this :)20:08
cjwatsonI understand that the GPLv3's system libraries exception is broader; I haven't checked how it would work in such a hypothetical situation20:09
cjwatsonGnuTLS could solve this using an LGPLv3/GPLv2 dual licence, if they wanted to: http://www.gnu.org/prep/maintain/maintain.html#Licensing-of-GNU-Packages20:10
pittinnnn20:43
StevenKmmmm20:43
pittioops, sorry20:43
* ScottK thought bug numbers were up to nnnnnn.20:43
StevenKSoon it will be nnnnnnn. :-(20:44
ScottKYep.20:44
pittiis there a prize for filing bug #1e6?20:44
ubottuLaunchpad bug 1 in Ubuntu "Microsoft has a majority market share" [Critical,In progress] https://launchpad.net/bugs/120:44
ScottKThat should be "on desktops"20:45
ScottKThey've got ~5% on smartphones.20:45
trismpitti: I think the --exec-prefix=/ is in the wrong spot in the dbus package rules, it is overriding the --prefix flag on the test build, so those binaries are overwriting the main ones (sorry if you're the wrong one to poke, I just notice you did the merges)20:50
pittitrism: not sure that I follow -- the package seems to be fine, it ships dbus-daemon and some tools in /bin, as it should?20:52
pitti--prefix /usr is actually intended, as we want manpages, pkgconfig, include files and whatnot in /usr20:53
cjwatsonthe claim earlier today was that /bin/dbus-daemon (IIRC) was from the debug build pass rather than the regular one20:54
stgraberpitti: around?20:55
trismpitti: if you look at the build log, the dbus-daemon is copied twice to debian/tmp/bin, once from build/bus and again from build-debug/bus20:56
mdzstgraber, kees, pitti, cjwatson, soren, TB in 520:57
trismpitti: so we get the one with all the DBUS_BUILD_TESTS sections enabled (the binary is twice the size of the one in debian)20:57
mdzish20:57
cjwatsonyep20:57
pittimdz: o/20:57
pittitrism: ah, I see20:58
pittitrism: have a meeting now, would you mind filing a bug about it? (or just mail me, but bug is nicer for tracking)20:58
trismpitti: sure, I'll file a bug20:58
pittitrism: thanks!20:59
=== and` is now known as and
=== and is now known as and`
=== JanC_ is now known as JanC
=== yofel_ is now known as yofel
=== bladernr_afk is now known as bladernr_
=== salem_ is now known as _salem
broderarges: backportpackage can't do that for local builds, but you could upload to a ppa - packages in a ppa can depend on other packages in a ppa22:43
argesbroder, awesome., i'll give that a try then23:04

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!