# /srv/irclogs.ubuntu.com/2013/04/04/#ubuntu-devel.txt

Bonez sarnold Bonez do any of you folks manage the actual ubuntu repositories? 03:38 some people here are archive admins.. 03:40 i'm assuming ubuntu uses dak to manage its repositories? 03:42 We do not. 03:42 Good morning 03:44 well that sucks. thanks anyways then. 03:44 ... 03:44 … indeed 03:49 === mnepton is now known as mneptok === jasoncwarner__ is now known as jasoncwarner_ === elky is now known as elky === Guest49821 is now known as vibhav === Ursinha is now known as Ursinha-afk === EntropyWorks_ is now known as EntropyWorks === SpamapS_ is now known as SpamapS === amitk is now known as amitk-afk === tkamppeter_ is now known as tkamppeter good morning 07:10 good morning Mr Holbach :) 07:10 salut didrocks 07:11 === smb is now known as smb === amitk-afk is now known as amitk === mthaddon` is now known as mthaddon === ckpringle_ is now known as ckpringle === Reanimator is now known as HerbertWest === doko_ is now known as doko cjwatson, is there light at the end of the haskell tunnel? 09:20 slangasek: due to a bug (https://bugs.launchpad.net/errors/+bug/1154189) I cannot see the graph for that bucket. 09:29 Launchpad bug 1154189 in Errors "Short URLs for buckets" [High,Triaged] 09:29 Hello 09:30 I have a branch to fix the URL issue. Please feel free to file a bug for anything amiss you find in the data (http://launchpad.net/errors/+filebug) 09:30 doko: it's some way off yet, but I think so.  Definitely making progress 09:42 jodh, should upstart-monitor be kept in main? 09:45 Daviey, zul: should cinder-backup, python-pybabel, quantum-lbaas-agent quantum-plugin-midonet  be kept in main? 09:46 doko: I believe so 09:47 seb128, didrocks: should gwibber, gcalctool be kept in main? 09:47 Daviey, could you seed these? 09:47 doko, no to both 09:47 doko: it's not essential, so maybe not. But would it be odd to have a single source package with binary packages in different repos? 09:47 doko, gcalctool is a dummy transitional package (got renamed gnome-calculator) 09:47 jodh, it's your call 09:48 jodh: No, that's not odd at all. It's quite common. 09:48 jodh: So don't let that affect your decision. 09:49 doko: ack 09:49 cjwatson, should libdrm2-udeb be kept in main? 09:49 seb128, can you help with account-plugin-icons? 09:50 doko, what about it? 09:50 should it be kept in main? 09:50 doko/soren: I'm easy where it lives then :) It just ended up in main by default. 09:50 doko, probably not, it's a dummy transitional package 09:51 seb128, ok, demoting 09:51 thanks 09:51 jodh, could you seed it? 09:51 doko: No.  Moved to universe 09:51 seed upstart-monitor> I would suggest lp:~ubuntu-core-dev/ubuntu-seeds/platform.raring supported-sysadmin-common 09:53 Or perhaps supported-sysadmin-desktop since it includes GUI support 09:53 (Doesn't matter that much, it's support lifetime pedantry) 09:54 cjwatson: thanks 09:55 seb128, should ubuntu-system-service be demoted? 09:55 or maybe removed? 09:56 seb128: ^ is that actually completely replaced by other things? didn't that set the apt proxy in the past? 10:01 doko, pitti: no, it's still useful and should still be installed by default 10:03 let me find the right place to add back a depends/recommends 10:03 what pitti said 10:03 seb128: did we switch over to system-services now? 10:03 ah, so we did 10:04 pitti, we did, and I dropped the u-s-s systemd compat parts 10:04 but as you said it still does stuff useful for us 10:04 we didn't get to move those bits somewhere else yet 10:04 seb128: what's left except the apt proxy stuff? 10:05 pitti, some keyboard stuff 10:06 pitti, but I'm not sure where/if we still use that 10:06 isn't that being done by accountsservice now? 10:06 another instance where I'd love to have codesearch.ubuntu.com :) 10:07 pitti, is has "is_reboot_required" as well 10:07 pitti, yeah, would be useful to have a codesearch, I'm wondering if we have anything left using those inferfaces 10:09 seb128: I guess apt proxy is g-c-c ? 10:09 pitti, it was, but we didn't port that feature to the new network panel afaik 10:09 so I'm not even sure it's used atm 10:09 uh 10:09 gnome-control-center-3.6.3/debian/patches/50_ubuntu_systemwide_proxy.patch:+                                       "com.ubuntu.SystemService", 10:10 we still have that 10:11 apparently upstream doesn't have a system-wide proxy config 10:11 pitti, series: "#50_ubuntu_systemwide_proxy.patch needs-rewrite" 10:11 it's not used at all in software-properties 10:12 pitti, right, one of the blueprint mention that it's planned to add to n-m 10:12 so I suppose the keyboard stuff can go as well 10:12 pitti, what's the right thing to grep for? com.ubuntu.SystemService I guess? 10:12 update-notifier and software-properties also don't use u-s-s 10:12 pitti, I will ask jdstrand if he can archive grep for it 10:12 seb128: *nod* 10:12 jdstrand, hey, is there any chance you could archive grep for com.ubuntu.SystemService in raring? 10:13 pitti, the only match in my local packaging dir is g-c-c and the patch is commented of the serie 10:14 so I suppose there's no need to add a dependency right now 10:14 i. e. raring just can't set system-wide proxy with a GUI 10:14 right, we might want to try to fix that before release though 10:15 not sure how important the feature is 10:15 * xnox had multiple reports from people not managing to get proxy settings right globally. E.g. apt-get works, yet downloading flash in the flashplugin-installer times-out and doesn't have the right proxy settings =( 10:25 is_reboot_required - shouldn't that now be handled by the upstart file bridge? or do we need to enable user sessions as well (not enabled by default for raring) ? 10:25 xnox, well, that code gives a dbus interface to query to know if a reboot is required 10:26 not sure what uses it, seems like that should be an info available in aptd though 10:27 that's useful for e.g update-manager or the indicator-session to turn the icon red 10:27 seb128: /me thought that we emit a dbus broadcast event upon flaggin up "reboot required" for the indicators and the like. They don't poll the status. 10:28 oh, they might on startup/first login. 10:28 right 10:28 stuff like update-manager are not running all the time 10:29 seb128: we can totally add that dbus interface to upstart, which has everything on dbus already. 10:29 seb128: e.g. a file-bridge dbus interface to check reboot_required file, wherever that file is. 10:29 wouldn't it make sense to have that interface provided by aptd rather? 10:30 that seems like the appropriate service for package management related apis 10:30 seb128: preferably we don't want to trigger aptd on every boot each time, though 10:31 it's really heavy 10:31 that's a good point 10:31 programs could just look for the stamp file 10:31 this is all heavily ubuntu specific anyway, so if we hardcode the ubuntu specific path or the ubuntu specific dbus name doesn't make much of a difference from a maintenance POV 10:32 but a huge one for performance 10:32 === mmrazik is now known as mmrazik|lunch the same is true for u-s-s, of course (being written in python) 10:33 === Sweetsha1k is now known as Sweetshark hey guys, i've just upgraded to 13.04 from 12.04, and now my single ext4 root filesystem doesn't unmount cleanly on reboot. a known bug? 10:48 pitti: looking at the code for bug 1103024 the __gsignal__ only specifies 3 parameters. Unless action-done is name clashing with some other "action-done" with 5 parameters? 10:50 bug 1103024 in apturl (Ubuntu Raring) "apturl-gtk crashed with TypeError in _on_finished(): 5 parameters needed for signal action-done; 3 given" [High,Confirmed] https://launchpad.net/bugs/1103024 10:50 === MacSlow is now known as MacSlow|lunch === ckpringle_ is now known as ckpringle === tjaalton_ is now known as tjaalton tkamppeter: build in the pipes 11:22 === michael is now known as Guest62910 === Guest62910 is now known as Mikch seb128: re archive grep> started 11:50 doko: some of the underlinking is reported during gtk-doc builds, which frankly we don't care about underlinking as it does build a scanner only to find relevant/wanted symbols. 11:51 can it be somehow disabled in gtk-doc? 11:51 jdstrand, hey, thanks! 11:51 === mmrazik|lunch is now known as mmrazik OdyX, OK. I am assuming that you are based on today's BZR commit (including the cups-browsed.postinst to overtake cupsd.conf entries). 12:03 === amitk is now known as amitk-afk jtaylor, I proposed a patch for bug 1164362, if it is accepted you can keep the versions in ipython's test control file. 12:08 bug 1164362 in autopkgtest (Ubuntu) " autopkgtest: doesn't support versioned dependencies" [Medium,Triaged] https://launchpad.net/bugs/1164362 12:08 jtaylor, just fix the deps for the test 'incomplete-install', there is a comma missing, and add python-lxml and python3-lxml to tools2 and tools3 dependencies respectively 12:08 === MacSlow|lunch is now known as MacSlow tkamppeter: saw them later on; yes. 12:11 tkamppeter: uploaded to experimental right now. 12:11 OdyX, thanks. 12:12 tkamppeter: I have pushed my changes to the repository. 12:13 tkamppeter: can you tag the repository for me. I'm done with fighting with bzr… 12:16 xnox, sorry, failing to parse that ... 12:35 didrocks: hi. upstream has approved my patch to Qt Creator fixing bug #1135336 , at  lp:~kubuntu-packagers/kubuntu-packaging/qtcreator 12:36 bug 1135336 in qtcreator (Ubuntu) "Qt Creator misconfigures itself on first run if qt4-qmake is installed (does not respect qt5-default)" [Undecided,In progress] https://launchpad.net/bugs/1135336 12:36 if you're willing to sponsor 12:36 === _salem is now known as salem_ doko: https://launchpadlibrarian.net/136066048/buildlog_ubuntu-raring-i386.glade-3_3.8.0-0ubuntu5_FAILEDTOBUILD.txt.gz fails during scanning docs. Can that "undefined reference to symbol" be optional at gtk-doc stage? 12:37 Mirv: oh, excellent news! do you mind summing that up for me on how this is fixed? 12:37 Mirv: like, I never lauchad qtcreator and installed qt4-qmake 12:37 what happens? :) 12:37 didrocks: it configures the Qt for you that is the default (Qt 5 in case of qt5-default) 12:38 Mirv: the fact to prefer qmake will should both qt5 and qt4 projects? 12:38 s/should/show/ 12:39 in the new project window? 12:39 didrocks: it will show both qt quick 1 and quick 2, if that's the question, yes 12:40 (Qt 5 supports qt quick 1 as well) 12:40 Mirv: exactly, thanks! :) 12:40 ah, that was what I meant, didn't realize the qt quick1 was coming from qt5 12:40 but yeah, makes sense 12:40 Mirv: building and sponsoring :) 12:40 xnox, I don't think so. so is this gtk-doc which needs the fix? apparently something is linked during the doc build 12:40 seb128, ^^^ 12:40 no idea about gtk-doc 12:41 didrocks: Qt Quick of Qt 4 was renamed to Qt Quick 1 when Qt 5 / Qt Quick 2 came out 12:41 Mirv: right right, I just didn't think about it when looking at the dialog that even qt quick 1 was indeed still coming from qt5 ;) 12:41 Mirv: nice work, the patch makes sense :) 12:41 thanks. nice that upstream agreed, I was wondering if they had some other logic on mind. 12:42 Mirv: yeah, it seems too obvious and tricky. Great that you get them reviewed first, I was afraid of side-effects :) 12:42 doko: so to generate gobject like API docs, a scanner is compiled and linked against a library one have just built. And like a lot of ugly things, it only links that .so and nothing else. As well frankly it doesn't care about linking in the dependencies. 12:43 doko: anyway I uploaded a fix in glade-3, but I wonder how many more of "failed to build gtk api docs" we have in that test rebuild. 12:44 xnox, seb128: well, you could patch gtk-doc to explicitly pass --copy-dt-needed-entries 12:45 -Wl,--copy-dt-needed-entries 12:45 doko: nice. well just patch the gtk-doc included makefile snippet to do that in the correct target. 12:46 (that snippet is used by all gtk-doc packages, as that's the way to build gtk-docs) 12:46 via include. 12:46 lamont, do you have a complete log for bug 1157687? 12:50 bug 1157687 in python2.7 (Ubuntu) "Upgrade to raring fails in prerm" [Undecided,Incomplete] https://launchpad.net/bugs/1157687 12:50 jibel, did you see this upgrade issue? ^^^ (or aren't you involved anymore with upgrade testing?) 12:59 doko, psivaa and plars are doing upgrade testing now. psivaa plars did you see this issue^ 13:02 doko: looking 13:03 jibel: no i did not see this issue, i normally update quantal before upgrading to raring and not sure if thats the reason why we do not see that. 13:05 xnox, are you working on the gtk-doc patch? 13:07 @pilot in 13:08 === udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Final Beta Freeze | Devel of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: ev doko: well back from lunch. Will quickly check how many other things failed at gtk-doc, and whether it's worthwhile to do in gtk-doc., 13:11 OdyX, why have you added your own init script for cups-browsed? there is an upstream one in utils/cups-browsed.in. 13:32 jibel, psivaa: no doesn't look like anything I saw either 13:36 doko: couldn't find another instance so far. 13:36 pitti: Do you know if/when there will be an update for postgres in the ubuntu repos for today's sec release? 13:36 nigelb: > #ubuntu-hardened maybe? 13:36 for security team. 13:37 xnox: Aha, thanks! 13:37 jibel, psivaa: only thing I had problems with was compiz (shocker!) :) 13:37 nigelb: way ahead of you :) 13:37 pitti: already building? 13:37 nigelb: it was built two days ago, mdeslaur is pushing out the announcements and packages right now 13:37 or already *built* 13:37 hah! 13:37 Good job you guys! :) 13:37 nigelb: upstream, debian security, ubuntu security and me are currently doing live coordination 13:38 * pitti currently feeding his backport PPA 13:38 \o/ 13:38 jdstrand: re the app launcher how is the app actually invoked such that this launcher is used 13:41 pmcgowan, the shell needs to make sure that the launcher is invoked as I understand it 13:41 doko: are you going to give back pycurl? 13:42 pmcgowan: that is TBD. we are going to write a prototype that is just an executable that can do 'applaunch /opt/../bin/foo' (or something) 13:42 pmcgowan: but we figured this would be more closely integrated with Unity 13:42 ok makes sense 13:43 pmcgowan: hence the 'handoff' portion 13:43 oh dear, the PPA autobuilders are totally blocked at a bad time 13:43 pitti: asking for score bumps would be justified 13:43 jdstrand: is the invocation just a middle-exec ? 13:43 cjwatson: I did bump them (psql security updates) 13:43 jdstrand, is there a work item for the apparmor definition step (leaving aside how to do it)? 13:43 or is it a service that exposes an API 13:44 === ricmm_ is now known as ricmm === timp is now known as t1mp tvoss: in that same blueprint, there is work for the apparmor definition. next month (iirc) we start looking at making it integrate with the SDK 13:45 jdstrand, cool 13:45 tkamppeter: which is too generic as far as I'm concerned. Also $cups and$alioth don't work. 13:46 jdstrand, we were initially thinking about an executor approach that allows for more fine-grained entry points to an application than just int main(...) 13:46 ricmm: my team is concerned really on with making sure the sandbox is setup properly. I think the Unity team is in the best position to define (with our review) how to execute a sandboxed app 13:46 I mean in terms of the launcher, just wondering what your planned implementation for it is in terms of consumers 13:46 pitti: hi, did you see debian bug 700838? do you want me to write a patch for it? :) 13:46 not just the definition of the sandbox itself 13:46 Debian bug 700838 in calibre "calibre: embedded copy of libjs-mathjax" [Normal,Open] http://bugs.debian.org/700838 13:46 mitya57: oh, please 13:46 ok 13:47 or maybe I'm reading this wrong and the launcher is actually a work item on someone elses plate, to match your sandbox definition 13:47 tvoss: sure. do you have existing code for this? 13:47 tkamppeter: also, it would be ideal if cups-browsed could 1) fork itself; 2) write its pidfile somewhere on his own. 13:47 jdstrand, nope, it was more an initial idea that would allow us to offer test executors and the like 13:48 ricmm: well that's just it. we don't have a planned implementation for consumers. we have a plan for the sandbox setup. we want to plugin to your (I'm assuming you are on the Unity team) plan for executing things 13:48 jdstrand, however, people vetoed the idea for multiple reasons, among them the lack of protection of the executor, and the argument that not every app will be an executable/shared library 13:49 which is why we are just doing a super simple prototype at first 13:49 tvoss: I'm not sure I understand that comment. can you rephrase? for an app to be confined properly, it needs to have its own pid for attachment such that we can attach a profile to it for that application 13:50 ie, multiple tabs for different HTML5 apps won't do 13:51 s/ie/eg/ 13:51 jdstrand, okay, the idea was that the executor is an executable that loads an app from a shared-object by resolving some pre-defined entry points, obviously, we could run the executor in your sandboxed env 13:51 (well, unless each tab is a separate process-- it gets slightly more complicated there, but it is technically doiable) 13:51 tvoss: this executor you refer to-- is that for the entire supported set of apps? (HTML5, QML, etc) 13:53 jdstrand, yup, that was the idea, although we didn't explore it any deeper tbh 13:53 pitti: unrar in dfsg tarball? 13:53 tvoss: ok, so your executor could be our app launcher (ie, we plugin to it to setup the sandboxed env) 13:54 tvoss: we have the aa_change_profile() library call that we can use in the executor to change into the apparmor profile at an appropriate time (eg, just before fork()) 13:55 jdstrand, interesting .. perhaps we stick to your first cut of main as an entry point and see how we can evolve things over time? 13:55 tvoss: I'm fine with proceeding how it makes sense. the prototype we want to have is really about showing how to setup the sandbox. that code can be pulled in to Unity in your executor or in some other fashion however it works best for Unity 13:57 pitti: If you have postgresql-9.1 ready for raring, go ahead and upload (or when you do).  I'm happy to go ahead and accept it, freeze or no. 13:57 jdstrand, cool :) that sounds like a really good idea 13:57 ScottK: ah, can do; I was going to sync it in about two hours or so, as dinstall is just running 13:58 jdstrand, can you point ricmm and me to the branch (under the assumption there is one already?) 13:58 tvoss: It seems practical to me to combine default handlers and apparmor setup in the same wrapper 13:58 pitti: OK.  That should be fine. 13:58 ScottK: but I can upload a fakesync now of course 13:58 tvoss: e.g. empty event handlers if the shared object doesn't provide this or that callback 13:58 tvoss: does this have any impact on making a qmlscene replacement for qml apps? 13:58 tvoss: cool :) so, we'd like to watch the executor work (coordinate, review, get involved), so if that is turning up on BPs, can you ping me and sbeattie (sbeattie will be working on the launcher) 13:59 ScottK: I'm going to do just that; let's cover all our bases 13:59 pitti: I'll trust you assessment of the urgency.  I mostly wanted to make sure you weren't waiting on beta 2 to release. 13:59 pmcgowan: We need one to set some env vars in any case 13:59 OK 13:59 albeit I'm not we have a list 13:59 *sure 13:59 tvoss: there is not one already 13:59 pmcgowan, totally not 13:59 vg 13:59 pmcgowan, think about it like a helper that just bootstraps a sandboxed env 13:59 sbeattie: ^ when you have something, can you point to tvoss and ricmm 13:59 jdstrand: tvoss lets entwine our names in single blueprint that contains the launcher/apparmor and executor story 13:59 from the point of view of the extended app lifecycle 13:59 not just shell->desktop 14:00 ricmm, +1 14:00 ScottK: oh, it's a merge anyway, darn; I didn't notice that we got ubuntu changes; uploading 14:00 Great.  Thanks. 14:00 ricmm: sure, we can move our prototype work items over to you-- we just did it where we did since we don't have anything yet :) 14:00 jdstrand: can we maybe have a call to explain how our lifecycle works right now? theres one thing that I deem important for this apparmor story and that is the resuming of previously-running applications after a power cycle and so on 14:00 we need to retore certain things, so all in all restore the sandbox... etc 14:01 === hggdh_ is now known as hggdh anyways, would be cool to have a voice call about it as sometimes chat can be slow 14:01 jdstrand: no I'm fine with confinement being handled by your team, I find it awesome 14:01 ricmm, do you think the app lifecycle bp makes sense? 14:01 great work indeed 14:01 tvoss: yes 14:01 ricmm: there shouldn't really be any concerns there-- apparmor policy is loaded into the kernel. any env variables are in memory and other stuff is filesystem, so a suspend/resume shouldn't need any particular attention 14:02 ricmm: but maybe I misunderstand your question 14:02 jdstrand: this is a power cycle scenario, apps exit memory 14:02 and we restore from saved states 14:02 I'd like to discuss the security concerns about it 14:02 and update our lifecycle drafts 14:02 ricmm: ok, please ask mdeslaur and jjohansen to attend the meeting 14:03 (I can also be there) 14:03 (mdeslaur is my techlead and jjohansen the apparmor lead) 14:03 awesome, thanks 14:03 ScottK: uploaded 14:03 tvoss: jdstrand any date preference? 14:03 ricmm, not on a weekend :) 14:04 jdstrand: is there a blueprint for your work somewhere? 14:05 so we can kind of bridge between our relevant ones and add them to the call 14:05 ricmm: this particular work item is in https://blueprints.launchpad.net/ubuntu/+spec/security-1304-appisolation-example 14:06 thanks 14:06 pitti: Thanks.  Accepted. 14:06 infinity, Daviey: ^ if you need to rebuild the server ISOs for beta 2 for some reason, including postgresql-9.1 9.1.9 would be a very good idea; I'll warn about this in my blog post 14:07 === fabo_ is now known as fabo === rickspencer3_ is now known as rickspencer3 ricmm: re date preference> none on my end other than what tvoss said :) 14:13 === mmrazik is now known as mmrazik|afk === security is now known as fire === lfaraone_ is now known as lfaraone === amitk-afk is now known as amitk\ === amitk\ is now known as amitk didrocks, ready? :) 14:40 dholbach: seems that rickspencer3 abandonned my hangout, so yeah ;) 14:40 hi didrocks 14:40 sorry 14:40 otp 14:40 :/ 14:40 no worry rickspencer3 :) 14:41 didrocks, we still have 20m - it's all good :) 14:41 === mmrazik|afk is now known as mmrazik Greetings. 15:24 ev: I was tolk I could bug you for an upload... :) 15:28 BigWhale: sure 15:30 well, depends on what you want to upload :) 15:30 ev: a new release of Kazam. This one, to be more preceise: https://launchpad.net/kazam/stable/1.4.2 15:32 1.4.1 is already in raring, this is just a minor bugfix release. 15:32 * ev looks 15:33 xnox: I'll look at that tomorrow; too much psql troubles this afternoon 15:36 OdyX, I had to release cups-filters 1.0.33, there was a NULL check needed. I have updated also the BZR for Debian. 15:39 BigWhale: is there a bug for this? 15:40 pitti: ack. 15:41 ev: yes, there were two... 15:42 They were targeted 15:43 BigWhale: numbers please :) 15:43 #1100626 #1085438 15:44 BigWhale: ah, I meant targeted against the Ubuntu source package 15:45 for the upload 15:45 since it will require freeze exceptions 15:45 BigWhale: see: https://wiki.ubuntu.com/FreezeExceptionProcess 15:45 Oh, none of them are. 15:46 BigWhale: would you mind creating one http://launchpad.net/ubuntu/+source/kazam/+filebug that requests the new version? 15:48 I'll then attach the debdiff to that 15:49 ev: https://bugs.launchpad.net/ubuntu/+source/kazam/+bug/1164584 15:51 Launchpad bug 1164584 in kazam (Ubuntu) "New version request - 1.4.2" [Undecided,New] 15:51 thanks! 15:52 * ev moves quick in his remaining 10% of battery 15:52 Oh, thank you. :) 15:52 on* 15:52 === deryck is now known as deryck[lunch] BigWhale: I've updated the bug with the debdiff and some follow-up instructions 15:59 ev: great, thanks! 16:00 once you have approval from the release team, let whomever is patch piloting know and they can shepherd the upload through 16:00 good luck! :) 16:00 @pilot out 16:03 === udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Final Beta Freeze | Devel of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: I just subscribed to a bunch of mailing lists... :> 16:05 === Ursinha-afk is now known as Ursinha ricmm: whoops, just got your invite... 16:15 ricmm: are you rescheduling it? 16:15 tkamppeter: ah, great, thanks. 16:32 mdeslaur: yes, rescheduling 16:40 Hey folks! I'm backporting a couple of python packages, A depends on B. I successfully backported B, but since it's only available in a PPA, I'm wondering how to indicate that the build for A should use that PPA to install B as a build-depends 17:05 === deryck[lunch] is now known as deryck roadmr: > better ask similar questions on #ubuntu-packaginging. But simply uploading into the same ppa will use depends from that ppa during build. 17:15 roadmr: well "biggest version wins", so packageB in ppa should be >> version number than in ubuntu archive proper. 17:15 xnox: oh perfect! thanks! and OK, I'll ask there next time, didn't know that channel existed :) 17:16 hey 17:25 xnox: so let's talk about command-not-found here 17:25 zyga: \o/ 17:25 xnox: so there are a few pieces 17:25 xnox: how to rebuild the data 17:25 xnox: when to rebuild it 17:25 xnox: when it should be documented 17:25 xnox: er, where 17:25 Hey folks! I'm backporting a couple of python packages, A depends on B. I successfully backported B, but si 17:25 xnox: and where it should be running 17:25 https://wiki.ubuntu.com/CommandNotFoundMagic 17:26 xnox: that's the original spec IIRC 17:26 with some *.pl URLs =))) ;-) 17:26 both are 404 though. 17:26 re 17:35 sorry 17:35 xnox: haah, I guess that's suxx.pl 17:35 xnox: sorry, need to move networks, brb 17:36 re 17:36 xnox: so let me see that page again 17:37 xnox: so the code is on launchpad which mirros github (or should, I'll update that) 17:37 xnox: the code is on github.com/zyga/command-not-found 17:37 xnox: inside the code there are a bunch of readme files 17:37 xnox: I should probably update the layout a bit to look less sucky 17:37 xnox: what do you expect in terms of automation 17:40 zyga_: at the moment, all I want is to run and update command-not-found-data. 17:41 zyga_: later i want instructions on a wiki page how to do that. 17:41 xnox: ok 17:42 xnox: so assuming you have the archive 17:42 xnox: you need to look at https://github.com/zyga/command-not-found/tree/master/UnifiedDataExtractor 17:42 xnox: and run ... 17:42 xnox: probably create-binary-database 17:43 xnox: let me check that in practice 17:43 === zyga_ is now known as zyga xnox: heh, it's been a while, let me do some basic cleanups to use it 17:48 pitti: I have a huge patch for you @ lp:~mitya57/debian/sid/calibre/remove-embedded-libraries 17:49 it removes bundled copies of mathjax, python-markdown and unrar 17:49 and also fixes some minor issues such as not working clean target 17:49 known issues: (1) it still has embedded jQuery 17:50 (2) it tries to write to /usr/share/applications/defaults.list during build 17:50 (3) the mime file is not installed 17:51 none of these seems to be caused by my changes though :) 17:51 * zyga cannot believe how crappy was his code just 8 years ago 17:52 zyga: to be honest python2.4 was a different kind of animal. 17:52 note that embedded unrar is a "Severity: serious" bug, but it doesn't affect wheezy 17:52 xnox: yeah but so were my python skills back then I guess :) 17:56 xnox: I think this was my first python program as well 17:57 xnox: and preinstalled on ubuntu, how cool is that :) 17:57 xnox: I didn't stress that much back then 17:57 xnox: the first commit is from 2006 though, I guess it's an import from something other than bzr, probably darcs 17:58 xnox: ok, so to run it 18:00 xnox: you need to run the scan script on the archive 18:00 xnox: basically on any debs you may have around 18:00 xnox: I've pushed a cleanup branch to git://github.com/zyga/command-not-found 18:01 xnox: I'll make usability nice 18:01 (enough) 18:01 xnox: do you think this should be a native package? 18:01 omg 18:02 whoever ran this 18:02 scan is still using /usr/bin/env python2.4 18:02 :DDD 18:02 haha :) 18:03 well, it probably runs on some arcane 10.04 systen 18:05 system 18:05 does 10.04 have python2.4? 18:05 xnox: perhaps there is a branch that is used by cannonical that is not in trunk somewhere? 18:05 Even 8.04 had python 2.5 18:06 wooow 18:07 I wonder when I really wrote that 18:07 did 4.04 had python2.4? 18:07 (times when python2.4 was new, OMG :) 18:07 plars: hey 18:08 roadmr: I kept making logging.debug("..." % (...)) mistakes back then :) 18:08 zyga: haha well the old you didn't have the present you to help with that ;) 18:08 haha 18:09 roadmr: anyway, even today, the code that implements get_alternatives in both crazy and amazing https://github.com/zyga/command-not-found/blob/master/UnifiedDataExtractor/scan#L41 18:10 roadmr: it actually tries to interpolate for loops and keep track of shell variable values 18:10 wow! reading 18:10 roadmr: it's not even close to correct but it improves results from 0% to something rather large 18:10 roadmr: I don't have the raw data to compare but when I wrote that, cnf is only usable with that hack 18:11 without it most of the things that people mistype/don't have around is not found as it is actually installed via alternatives system 18:11 well 18:13 back then mirroring the archive was hard 18:13 I remember it filled my _spare_ disk 18:13 and I took a week to download all the packages 18:13 I guess I can start the mirror now, put my kids to bed and have the full archive after that 18:14 it was a different time, before fiber was available everywhere 18:14 dobey: yeah 18:22 dobey: note: I'm still on ADSL, no fiber around where I live 18:22 dobey: but it's 25MBit, not 0.1 18:22 wow heh 18:23 I guess I can make it a bit better now 18:25 like scanning partner repo 18:26 and getting skype :) 18:26 and scanning arm 18:26 wow 18:26 that's really old stuff there 18:26 xnox: I'll get you the data for raring in a few hours 18:26 xnox: what's the total deadline for this? 18:26 zyga: between now and 18th of April. 18:28 the updated command-not-found should be uploaded. 18:28 ricmm: hey, I'd like to attend the meeting you scheduled for tomorrow, but I have a conflict. it is possible to move it a half an hour earlier? 18:29 s/it is/is it/ 18:29 xnox: ok 18:30 xnox: while I'll clean the code a bit we can just upload the data file 18:31 xnox: do you know where are the armhf files in the archive? 18:32 xnox: http://archive.ubuntu.com/ubuntu/dists/raring/ looking there I can only see i386 and amd64 18:32 zyga: ports 18:33 zyga: http://ports.ubuntu.com/dists//main/ 18:34 ah 18:34 ok 18:34 zyga: https://wiki.ubuntu.com/SecurityTeam/FAQ#Architectures 18:34 thanks 18:34 pitti: filed a MP (https://code.launchpad.net/~mitya57/debian/sid/calibre/remove-embedded-libraries/+merge/157195), have to go now 18:34 * zyga starts mirroring the archive, I'll be back later when that is finished 18:36 === kees_ is now known as kees rickspencer3_: ping 19:37 xnox: ping 20:50 xnox: quick question, let's say I'm going to build a manually-verified database that annotates _all_ ubuntu packages with the update-alternative targets that each particular package might or does use; I could then automatically annotate the control files of all of those source packages with the appropriate meta-data, could we then merge this (big) delta back into debian so that we don't have to maintain it and so that command-not-found gets reliable, perfect 20:51 slangasek: ^^ (maybe you can also say how crazy I am :) 20:52 xnox: I'll start with a standalone database (text, editable) that keeps each package-version -> mapping of update-alternatives used in version control 20:53 xnox: then try to get the latest version of all those packages in source form, patch the control file, generate a diff from that 20:53 xnox: do you think that's the right approch to getting that information back to debian? 20:53 zyga: I'm slightly confused by above. In general we already have debtags on debian side for "annotating" packages. As all of this data gets volatile very quickly. 20:55 xnox: could you quickly describe those debtags? 20:56 xnox: are they in the source package or somewhere on the side? 20:56 xnox: (I remember this discussion years ago, I remember someone said 'ah yes, debtags can solve that', but debtags were still under discussion back then) 20:56 xnox: my goal is to have _reliable_ meta-data as to what update-alternatives targets each package uses 20:57 xnox: that data is required for command-not-found 20:57 zyga: the general ideas about such things tends to end up in "AppStream" proposals (multiple incompatible with same name but different implementations) such that package self annotate a bit, but archive publishing software (dak in debian, launchpad in ubuntu) further scans source/binary packages and publishes one co-herent all-purpose index. 20:57 xnox: instead of mainitaining this data separately I'd like to get it back to debian 20:57 neither of which are at a usable state at the moment. 20:57 zyga: http://debtags.debian.net/search/?q=game 20:57 xnox: this cannot be automated 20:58 xnox: it's the same kind of data as Depends or Build-Depends 20:58 zyga: sure. with AppStream proposal parts that cannot be automated live inside the package (source or binary as appropriate) other parts which can be automated live in archive software which merges the two together. 20:58 xnox: I see that, quickly and ugily as X-Update-Alternatives-Targets: /usr/bin/vim.real=/usr/bin/vim 20:59 xnox: so I don't care about the automated part simply because it's not automated in any way at all 20:59 xnox: how does one express the non-automated variant of that in the control file? 20:59 zyga: the beauty of update-alternatives is that we have packages from 3rd parties hooking into that, e.g. like google-chrome web-browser and packages comming from outside of the archive (e.g. things like extra/new versions of emacs, postgresql etc) 21:00 xnox: great 21:00 xnox: so what? 21:00 xnox: for high-profile packages we can keep side-data in command-not-found itself 21:01 xnox: for the archive, I'd like to keep this _in_ the archive, otherwise it's something that's forever broken 21:01 xnox: it's something that packages would have to maintain 21:02 xnox: the benefit is accurate hit database from commmand-not-found 21:02 xnox: hint 21:02 xnox: I'm willing to do the initial scan myself, manually 21:03 zyga: i think packages will ok self declaring "i update-alternative emacs with vim.real binary I provide", and then one can quickly scan and collate all those that declare "i update-alternative emacs" 21:03 xnox: I'm just looking for a way to contribute this back 21:03 xnox: packages or packages? 21:04 zyga: it's best to raise that on debian-devel with a sample scan results & carefully explaining what you are trying to achieve. Prepare for a flamewar of being shot down with "apt-archive Package index is too big already" and "we don't need this" and "one day when dak supports this we may consider it" 21:05 zyga: but hey maybe there are other people who where looking to achieve this as well. 21:05 xnox: the last item is exactly what I want to avoid 21:05 zyga: plan your email well =) 21:06 xnox: I want to get this done done for ubuntu 21:06 xnox: getting this to debian is a bonus 21:06 xnox: sure 21:06 zyga: ubuntu-devel then =) 21:06 xnox: I'll get the initial database today/tomorrow 21:06 xnox: do you think I should discuss this in ubuntu over debian first? 21:06 xnox: given my history of collaboration with debian, I'd rather not start wrong again 21:07 zyga: the amount of overlap between the two mailing list is ever increasing =) no harm imho to start it on either lists & bounce to the other mailing list after initial rough corners are considered. 21:08 ok 21:09 xnox: I'll start with ubuntu-devel to get the c-n-f goal done 21:11 ubuntu/debian defaults to IET while Redhat defaults TGT, openstack/cinder now defaults to TGT due to ubuntu's efforts, confused 21:29 t1mp, hi 21:32 rickspencer3: hi, I saw your bug report https://bugs.launchpad.net/ubuntu-qtcreator-plugins/+bug/1161910 and created an MR that should fix it. 21:41 Launchpad bug 1161910 in Ubuntu QtCreator Plugins "creating a single page app adds a weird bar to the top of the screen" [High,Confirmed] 21:41 rickspencer3: it is eod for me, but feel free to try out my changes and comment on the MR 21:41 thanks t1mp, let me know if you want me to test it in some way 21:41 t1mp, will do 21:42 rickspencer3: thanks. You didn't include code with the bug, but if there is still some weird positioning happening, make sure that you do not set the anchors of the Page (that should be automatic) 21:43 t1mp, well, it was a defualt app 21:43 so, it should be easy for me to test it out 21:44 ogasawara, the gcc-4.7.3 release candidate is currently building in raring. the version is now set to 4.7.3, just in case that you still depend on that 21:44 jtaylor, why the syncs of the new java lib packages? 21:49 doko: bugfix releases 21:49 1.1.0? 21:50 yes I was confused to 21:50 but the changes lists only bugfixes, diff is also small 21:50 I was asked by the debian maintainer to sync them 21:51 === kentb is now known as kentb-out === salem_ is now known as _salem === Ursinha_ is now known as Ursinha === rsalveti_ is now known as rsalveti

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