[00:20] hi everyone. anyone seen utlemming or anyone else responsible for the AWS AMIs around? === mako_ is now known as mako === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha === jamesh_ is now known as jamesh [05:01] Good morning [05:02] xnox: ack, thanks; we have that in Steve's bug reply as a reminder [06:54] good morning [06:59] morning dholbach :) [06:59] salut didrocks [07:31] RAOF: oh dear, I was blaming the new libgphoto2 all along, but it seems the @DEV stuff broke umockdev recording [07:31] RAOF: and on top of that, @DEV isn't actually written; I'm looking at both [07:32] pitti: Huh, for what? I rather thought I'd *used* that feature! [07:32] RAOF: just FYI in case you try to actually use that with Mir [07:32] I, um, *have* used that in Mir. [07:32] RAOF: if you record several commands (or more generally, open/close cycles of the device), only the last is kept [07:32] instead of all of them [07:32] Oh. I don't do that. [07:33] So wouldn't have noticed. [07:33] RAOF: but the way how ioctl records work the @DEV was never written to them [07:33] RAOF: oh, unless your program never called close() but you just ^Ced it [07:33] Ding! [07:33] which would be with evtest === smb` is now known as smb [07:47] infinity: nice job on bug 1319122! [07:47] bug 1319122 in tzdata (Ubuntu Trusty) "tzdata needs expedited SRU for Egypt DST change on May 15" [Undecided,Fix released] https://launchpad.net/bugs/1319122 [07:53] RAOF: hm, still a miracle how it worked with ^C though, as it essentially does the same thing as with close() [08:10] pitti: It's also possible that I hand-edited in the @DEV line because I was umockdev-recording with the system-wide binary which didn't yet have the patches. [08:11] RAOF: ah, that sounds plausible then [08:11] RAOF: anyway, fixed now, releasing 0.8.2 [08:13] pitti: Yeah, I don't have high hopes that many computers actually got updated in time, but at least we did our best in Debian/Ubuntu. [08:14] infinity: one thing that's weird is that utopic has the trusty SRU; we can't sync from Debian? [08:14] pitti: Look again. [08:14] pitti: I just copied the trusty SRU over as a stopgap while LP caught up with dinstall. They're in sync now. [08:14] ah :) [08:14] thanks [08:16] pitti: It was just a complete fluke that an Egyptian user popped into #debian-glibc to tell us about it, mind you. I might suggest that Paul Eggert set up an emergency broadcast list for these rare times when governments go crazy with 0 notice. :P [08:17] I thought there was some tz-announce@ thingy [08:17] If there is, no one seems to be on it. [08:17] There sure is... [08:18] Oh, but that's just his release announces. [08:18] I should subscribe to that anyway. [08:18] But I meant something even lower traffic and higher visibility for "OMG THIS COUNTRY IS CRAAAAZY, UPDATE NOW!" [08:19] so what was Argentina back then is now Egypt :) [08:19] In this case, though, thanks to said user, I think we got all the updates out within hours of Paul's release. [08:20] There, subscribed to tz-announce, at least. [08:39] hum [08:39] what's going on with e.u.c? [08:40] https://errors.ubuntu.com/?period=day looks bugg [08:40] ev, bdmurray: ^ do you know if there is a known issue? [08:40] +y [08:43] seb128, bdmurray: that's troubling. Brian, do we have any recent changes to production that could've caused this? I had a look through RT and the daisy and errors trunks, but I don't see anything that landed in the past day or two. [08:44] ev, I guess Brian is not going to be up before some hours [08:44] yeah [08:44] on first blush it looked like it was only showing package installation crashes, but there's at least one regular crash in there [08:45] well, default view is the packages you are subscribed to no? [08:46] it gives me an empty list here [08:46] (well, now it's stucked on "Loading...") [08:47] but like if I do trusty reports on 1 day [08:47] I get "no data to display" [08:47] https://errors.ubuntu.com/?release=Ubuntu%2014.04&period=day [08:57] https://graphite.engineering.canonical.com/dashboard/#15may-crashes suggests that we are at least capturing the data [09:08] ev, do you want me to file a report about the issue? against what tracker/component? [09:13] seb128: please, against lp:errors [09:14] I *think* (/hope) at this point that the problem is in the UI layer, not in our data collection [09:24] ev, https://bugs.launchpad.net/errors/+bug/1319730 [09:24] Launchpad bug 1319730 in Errors "the current reports are (mostly) empty" [Undecided,New] [09:32] please how do i build a software on ubuntu [09:36] ubilli8, depends [09:36] seb128: thanks! ^ bdmurray would you mind digging at that when you get a chance? It doesn't appear like we're dropping data on the floor in daisy, but please do confirm in case of I've missed something obvious :) [09:36] ubilli8, which software [09:37] ev, bdmurray: thanks for looking at it ;-) [09:40] i want to develop LAMP just like WAMP for windows but i want to know how the ubuntu architecture works to do that and what language to use to write the language [09:43] @ [09:43] Error: "" is not a valid command. [09:44] ubilli8, 'LAMP' is not one piece of software [09:45] yeah something like wamp that holds all the software during installation and manages it. just like wamp [09:46] ubilli8: Are you referring to http://www.wampserver.com ? [09:47] ubilli8, so some sort of gui for managing the installation? [09:48] yes do you think it is possible...??? [09:48] ubilli8: Anyhow, I'll throw you a bone. To replicate that set of packages installed, you want "apt-get install lamp-server^" and "apt-get install phpmyadmin", but this is also a question for #ubuntu, not #ubuntu-devel, this isn't a channel for people who develop using Ubuntu, but for people who develop Ubuntu itself. [09:50] xnox, why would you not ship the cmake tools sdk-libs-dev ? cmake itself is shipped there too ... sounds kind of logical to me to ship the needed tools for development in there as well [09:50] (i would have asked you in #ubuntu-touch ... but you seem to not be there) [09:51] ogra_: hm? i explicitly said that they should be shipped in sdk-libs-dev. [09:51] * ogra_ re-reads the mail [09:51] ogra_: but since it's a metapackage -> that means in one of the seeded depedencies. [09:51] ogra_: metapackages should be empty. [09:52] ogra_: especially those generated from seeds/germinate. [09:52] xnox, heh, lol, ignore me ... you said exactly what i meant [09:52] i totally misread [10:34] what is the *correct* way to disable all i18n aspects? so far I'm using LANG= LANGUAGE= LC_ALL=C.UTF-8 but I have no real reference that would say this is correct and minimal, setlocale(3) is not really helping much [10:35] from what I see, setting LANG=C is insufficient, I get a mixture of translated/localized text, equally insufficient is LC_ALL [10:35] is there a document that says how l10n is supposed to be initialized and specifically, disabled [10:36] LC_ALL=C [10:37] or LC_ALL=C.UTF-8 if what you actually mean is "disable locale-specific things but let me have UTF-8 character encoding" [10:37] (note, C.UTF-8 is a bit system-specific, but it's always there on Debian and descendants) [10:37] actually, sorry, you must also unset LANGUAGE because LC_ALL doesn't imply that [10:38] cjwatson: yeah, pure LC_ALL is irrelevant it seems [10:38] but LC_ALL overrides LANG and all of LC_* so you don't need to deal with those separately [10:38] LC_ALL is NOT irrelevant [10:38] http://paste.ubuntu.com/7467208/ [10:38] it's just not quite sufficient [10:38] LC_ALL=C.UTF-8 pactl list [10:38] LANGUAGE= LC_ALL=C.UTF-8 will be sufficient though [10:38] right, it doesn't do much effectively though (I get it that it gets respected though I don't understand why it's not overriding LC_MESSAGES [10:38] no, it just doesn't do one specific thing [10:38] cjwatson: which thing is that? [10:39] cjwatson: is LANGUAGE documented anywhere? [10:39] info libc [10:39] LANGUAGE overrides LC_ALL IIRC just for the purpose of the message translation category [10:40] cjwatson: thanks [10:40] but that's just for gettext; for all other locale categories the master variable is LC_ALL [10:41] cjwatson: reading the relevant part of the info page [10:42] so for instance collation order for sorting things, case conversions, numeric/monetary/time presentation, etc. [10:42] cjwatson: hmm, I think there's a bug though still [10:42] cjwatson: look at gettext(3) [10:42] cjwatson: LANGUAGE is apparently ignored if locale is "C" but apparently it doesn't know about C.UTF-8 which is for all intents and purpuses just better "C" [10:43] I expect that would be worth fixing in gettext(3), yes [10:43] I suggest filing a bug, I don't work on this stuff :) [10:43] * zyga just verified that C is special cased and C.UTF-8 isn't [10:43] cjwatson: yeah, I'll file a bug on that [10:43] cjwatson: thanks for showing me info libc :) [10:44] yw [10:45] ara: hey, your office space fixed their internet filters :) [10:46] zyga, no, I just moved back home :D [10:47] zyga: Can you file a Debian bug against src:eglibc for the lack of special-casing of C.UTF-8 there? [10:47] infinity: my pleasure [10:47] zyga: Ta. [10:48] This is likely a missing piece I need to fix before I start enacting my "C.UTF-8 everywhere" plans. [10:48] arges: yes, rsync the rootfs and also copy over the files under $lxcpath/$name, i.e. /var/lib/lxc/u1/config and if it exists .../fstab [10:49] infinity: :) [10:49] infinity: I could patch it, should be relatively simple [10:49] zyga: Yeah, I didn't assume it would be hard, but since you've already gone and found the relevant bit, a bug with a pointer would be awesome, so I don't duplicate the effort next week. ;) [10:50] * infinity wonders if maybe it's about time to try to push C.UTF-8 upstream for glibc 2.20, and puts that on his "talk to aurelien" list for when it's not 5am. [10:51] ok, uploading lts stack part 1 (all !drivers) === MacSlow is now known as MacSlow|lunch [10:54] gah, reportbug just ate my bug description [10:57] infinity: sent, I'll give you the number as soon as I get it back [11:00] Speaking of C.UTF-8, can anyone reproduce https://errors.ubuntu.com/problem/70c1930688440b1818145db3346da61eb39a7ac8 ? [11:02] infinity: on 14.{04,10}? no [11:03] zyga: Either. [11:04] Seems to work fine for me at a python3.4 interactive prompt. No idea why it's crashing there. [11:04] * infinity shrugs. [11:05] infinity: I've seen odd locale crashes but typically mid-upgrade or remotely [11:05] infinity: but software properties [11:05] wait, could that be add-apt-repository? [11:07] infinity: might be worth to see on a cloud image, those are usually locale-challenged [11:08] infinity: debian bug 748215 [11:08] Debian bug 748215 in src:eglibc "gettext(3) should special case both C and C.UTF-8 wrt LANGUAGE lookup" [Wishlist,Open] http://bugs.debian.org/748215 [11:09] zyga: Ta. [11:10] locale-challenged> they are, but C.UTF-8 is in libc-bin, it should be everywhere [11:12] cjwatson: true [11:12] cjwatson: does it need to be generated somehow though, or is it always available? [11:13] zyga: It's always available. [11:13] zyga: That's why that crash confuses the heck out of me. [11:13] infinity: yeah, I can understand that [11:13] (And why I can't reproduce it in a minimal test case in a barebones chroot...) [11:13] infinity: I have no idea why it might occurr [11:13] It *is* possible to remove it, but you have to ignore the "you're trying to remove an Essential package, you idiot" prompt [11:13] Maybe somebody did [11:13] cjwatson: ;-) [11:13] mvo worked around it by just trapping the error, but it shouldn't error in the first place. [11:14] cjwatson: A fair few someones... [11:14] infinity: the random instance I looked at had a freshly installed 14.04 (0 days old) [11:14] infinity: so I doubt they would remove those packages really [11:14] Unfortunately libc-bin isn't in Dependencies (since it's Essential) [11:15] Yeah. Oh well. Unless someone can reproduce it and tell me how, it'll be one of those things that just annoys me peripherally but I won't do anything about. [11:15] I doubt anyone would remove libc-bin, but I suppose I could see someone deleting the locale itself. [11:15] But still, that's a lot of crash reports for something as silly as that that I can't see many people doing. [11:15] "we don't need this bit" [11:15] yeah, as you say [11:16] right, I'm also quite puzzled by this fwiw [11:16] We do have a race in locale-gen that I intend to fix, which could actually cause this sort of issue in the middle of a libc6 upgrade, but since fresh installs see it, and I've not SRUed glibc yet, that theory's out the window. [11:17] Oh, and c.utf-8 is pre-generated in the package anyway. [11:17] So... I dunno. [11:17] I shall try to forget it, like so many bad relationships^wbugs passed. === _salem is now known as salem_ === MacSlow|lunch is now known as MacSlow === psivaa is now known as psivaa-afk === greyback is now known as greyback|food === tedg is now known as ted === ted is now known as tedg === doko_ is now known as doko === greyback|food is now known as greyback [13:30] is any one here? === psivaa-afk is now known as psivaa === pete-woods is now known as pete-woods-lunch [13:57] xnox: hm, the new sysvinit seems to have broken at least LXC, and that looks real (i. e. not just like a CI machinery quirk) [13:57] pitti: ouch, i've tested everything extensively. [13:57] pitti: how/what ? [13:58] Setting up lxc (1.0.3-0ubuntu3) ... [13:58] lxc start/running [13:58] invoke-rc.d: unknown initscript, /etc/init.d/lxc-instance not found. [13:58] http://d-jenkins.ubuntu-ci:8080/job/utopic-adt-lxc/22/ARCH=i386,label=adt/console [13:58] or https://jenkins.qa.ubuntu.com/job/utopic-adt-lxc/22/ARCH=i386,label=adt/console for folks without VPN [14:00] pitti: [14:00] $ sudo initctl --system status lxc-instance [14:00] initctl: Unknown parameter: NAME [14:00] Usage: NAME=name of LXC instance [14:00] I have a /etc/init/lxc-instance.conf, but not /etc/init.d/lxc-instance [14:01] + && initctl status ${SERVICE} 2>/dev/null 1>/dev/null [14:01] oh, likely due to that? [14:02] xnox: right, you're ahead of me [14:02] so apparently status doesn't work for these "template" jobs [14:02] pitti: it think it's a bug in lxc postinst, it should not have "dh_installinit --name lxc-instance --no-start -p lxc" [14:03] pitti: as there is no way to invoke lxc-instance without an instance.... [14:03] pitti: verifying that now. [14:13] pitti: hm, that is kind of sad. But e.g. initctl show-config still lists lxc-instance. [14:14] xnox: you (or Debian, not sure) does this instead of a simple [ -e /etc/init/lxc-instance.conf ] to also catch overrides etc.? [14:14] pitti: yes. [14:14] pitti: and multiple configuration directories, as we will need in the future. [14:15] pitti: it's more strict, but more correct at the same time. And e.g. i'll be able to resurrect php upstart job. [14:15] xnox: so perhaps only check if "LC_ALL=C initctl status $job | grep -q Unknown"? [14:15] unfortunatley the exit status always seems to be 1 [14:16] i. e. both for the lxc-instance "template" situation and a real unknown job [14:16] (upstart job added in trusty, with syntax not supported by precise upstart. and at the moment invoke-rc.d tries to use upstart job, instead of the init-script whilst pid1 is still precise-upstart which has no clue about this new job with what it thinks is invalid syntax) [14:17] pitti: i'll check other template jobs, but for lxc it really should be doing --no-start in it's dh_installinit calls. [14:17] ah right, it doesn't [14:19] pitti: at least no pam-systemd should configure in the lxc container just fine, even if one does lxc-attach to it =) [14:19] pitti: will email people about it, once verified. [14:19] xnox: nice! [14:19] s/no/now/ [14:20] Hi, is it possible to disable building ddebs in sbuild? From PPA build logs, there is a line that seems to does this. (dh_strip debug symbol extraction: disabling for PPA build) [14:21] it can be enabled/disabled by PPA [14:21] saiarcot895: but that means it's already not happening, i. e. it shoudln't spit out ddebs [14:21] saiarcot895: or in packaging you can export a variable to make sure pkg binary mangler doesn't create ddebs. [14:21] pitti: The PPA isn't spitting out ddebs, but sbuild locally is spitting out ddebs. I'm trying to disable that [14:22] saiarcot895: ah; drop the pkg-create-dbgsym package from it then [14:22] pitti: could you have a look at bug 1318034? [14:22] bug 1318034 in Daisy "indicator-sound crashes frequently fail to retrace" [Undecided,New] https://launchpad.net/bugs/1318034 [14:22] saiarcot895: you probably have binary pkg mangler installed. [14:22] saiarcot895: i have this in my ~?.sbuildrc [14:22] saiarcot895: $build_environment = { 'NO_PKG_MANGLE' => '1', 'DEB_BUILD_OPTIONS' => 'parallel=12', 'HOME' => '/build/' }; [14:22] bdmurray: yes, will do [14:22] saiarcot895: you want the NO_PKG_MANGLE => 1 option. [14:23] xnox: I'll try that [14:23] that should work too, yes [14:45] pitti: i uploaded fixed lxc, and i hope in time autopackagetest will migrate both. [14:46] xnox: http://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html#sysvinit is looking good so far [14:46] xnox: and I'm glad this stuff is now actually useful, now that britney is fixed :) [14:50] pitti: looks like that's the only package affected so far. Well at least on my system, didn't check the $world. [14:51] xnox: so grep -r ^instance /etc/init/* are all potentially affected ones, right? [14:51] pitti: well & calling invoke-rc.d in their postinst. [14:51] pitti: for all those that i had on my machine, there are none that called invoke-rc.d. [14:52] pitti: well none, apart from the now fixed lxc-instance. [14:52] nice [14:53] stgraber: i did direct to archive upload of lxc, packaging change only. Not sure if it needs to be committed somewhere else as well. [15:00] I figure at least to Debian? [15:01] pitti: packaging appears to be a fork. and in debian --no-start is passed. [15:02] ah :) [15:14] mvo: Would you mind retargeting your click merges to lp:click/devel? I've just created that. [15:16] cjwatson: sure, no problem [15:22] xnox: let's talk about py2 [15:22] barry: > #ubuntu-ci-eng [15:23] barry: oh, you are not there. I was chatting to ogra_ about it there already. [15:24] xnox: i'm there now [15:31] xnox: http://d-jenkins.ubuntu-ci:8080/job/utopic-adt-lxc/23/ARCH=i386,label=adt/console crashed with something weird in apt-get update, retrying [15:31] xnox: but amd64 succeeded === pete-woods-lunch is now known as pete-woods [17:53] hey I'm trying to copy a package to my PPA and I'm getting "source contains expired files". what's the deal with that? [17:54] I went spelunking into the superceded section of old libav builds and found one that I want to put into my PPA for precise [17:55] and launchpad won't let me do it [17:55] hmm maybe I should go to #ubuntu-app-devel === roadmr is now known as roadmr_afk === tedg is now known as ted [19:08] arges: are looking at the trusty SRU queue? [19:09] bdmurray: had a few minutes between a test I was running. sorry if I steps on your toes a bit [19:09] arges: not at all, I was just gonna get started [19:09] bdmurray: ok cool : ) === roadmr_afk is now known as roadmr [19:45] xnox: I'm used to pick it up when changing in the archive, that's fine [20:21] arges: bug 1316125 is fixed in trusty presumably? [20:21] bug 1316125 in autofs5 (Ubuntu Saucy) "Autofs leak file descriptors when reloaded (-HUP) and daemon may stop working on high # of shares/reloads" [Undecided,In progress] https://launchpad.net/bugs/1316125 [20:22] bdmurray: let me verify [20:23] arges: it doesn't look like it to me... [20:28] bdmurray: yup il'l have tinoco fix that thanks [20:29] yep, im on it [20:29] arges: okay, I've also added an autofs task since the closure looks at the source package name [20:29] oh didn't realize it was targeted against the binary package [20:29] bdmurray: thanks [20:30] I think there was some weird renaming of autofs / autofs5 [20:31] and if its not fixed in trusty then it isn't fixed in utopic since they are the same verison [20:31] bdmurray: the same version was kept from saucy to utopic [20:32] and this bug fix is based on a commit on 5.0.7 so all of them needs the fix [20:32] im creating the diffs for them [20:32] i'll attach to ua and lp, ok ? [20:50] kirkland, hey, shift+f2 stopped working a day or two ago in my byobu under utopic (shift+f7 stopped working before that), both print ~ at that point, any idea where that might come from? [21:34] can one add new source packages to trusty? [21:36] through a backport [21:36] I mean into updates [21:37] so other srus can depend on it [21:37] does an important bug fix depend on a new source package? o_O [21:37] yes [21:37] go on [21:37] one could do it in an existing one but thats incredibly ugly [22:10] robert_ancell, when were you thinking of releasing the next lightdm revision? [22:11] mterry, whenever, is there something you were blocking on? [22:11] robert_ancell, split greeter would appreciate that login1-race fix [22:11] mkay === Ursinha is now known as Ursinha-afk [22:14] mterry, perhaps you could also look at https://code.launchpad.net/~christian-w/lightdm/qt-binding-keyboard-layouts/+merge/205356 [22:15] jtaylor: Yes, but it's rather ugly and needs some justification. [22:16] robert_ancell, looking [22:17] justification is I don't want to touch tcl ._. [22:18] jtaylor: try again =) what's wrong with tcl in trusty? we even have two, if i recall correctly [22:18] thats the problem [22:19] I want to add a new package using tcl 8.5 instead of 8.6 [22:19] to make the stuff we didn't transition work again [22:19] jtaylor: you don't have to build against both, you can use just one of them. and both should be in main. [22:19] so far I know porting is either non-trivial or impossible [22:19] xnox: doesn't work with libraries [22:20] one library uses 8.5 the other 8.6 [22:20] you can imagine that that will not end well [22:20] jtaylor: what's the incompatible set of packages you are talking about? [22:20] the blt and itcl3 rdepends [22:20] it seems itcl3 was incorporated into tcl 8.6 as itcl4/tclOOP [22:21] I have not found if you can even use itcl3 with 8.6 [22:21] trying it just crashes [22:21] jtaylor: so you are saying that skycat & tkdesk are borked? [22:21] yes [22:21] jtaylor: well one or the other way it needs to be fixed in utopic, and is probably worth an sru into trusty. [22:22] jtaylor: have you filed a bug yet? [22:22] I would do that by adding a blt8.5 [22:22] which is really easy [22:22] as its library is versioned [22:22] the question is can I do that in trusty too [22:23] jtaylor: you don't need to add a new source package though. you can just add a second binary package. [22:23] xnox: I know but thats not easy [22:23] jtaylor: or it would be best to resolve it by e.g. rebuilding itcl3 with 8.5 in trusty. but that needs testing. [22:23] itcl is 8.5 in trusty [22:24] blt would need rebuild against 8.5 [22:24] well then blt should be to. [22:24] right, that. [22:24] which is problematic as its rdepends use 8.6 === Ursinha-afk is now known as Ursinha [22:24] its safer to add a new package [22:24] given that blt's rdependes is relatively short, we will probably need to rebuild those as well. [22:24] reverting a transition seems more ugly than adding a new source [22:25] jtaylor: a new package doesn't solve the problem that you have binaries linked together with incompatible tcl's and only adds more cases to the depedency chart. [22:25] it helps at least for skycat [22:25] well in utopic we really should port itcl3 to 8.6 [22:26] and in the best interest of everyone cherrypick that into trusty..... [22:26] jtaylor: please open a bug report against blt, itcl3, skycat & tkdesk. [22:26] jtaylor: we'll need to track this, whichever way this is going to be solved. [22:29] I don't know how to port it [22:29] I can't even get a reasonable backtrace of the crash ._. [22:30] jtaylor: that's ok, others know how to port things, but please open the bug report describing the issues you are seeing. [23:00] mdeslaur: Re bug 1319970, I am happy to take care of this if you like. I have commit access to the git repo for speech-dispatcher in Debian, and I track Ubuntu's changes there too. [23:00] bug 1319970 in speech-dispatcher (Ubuntu Utopic) "speed-dispatcher user needs a restricted shell (/usr/sbin/nologin or /bin/false) instead of /bin/sh." [Undecided,Confirmed] https://launchpad.net/bugs/1319970 [23:00] When building Qt 4.8.5 on Trusty, there's a message printed out about using the -no-exceptions flag (http://paste.ubuntu.com/7470347/). Might this be beneficial? [23:07] mterry, 1.11.2 uploaded [23:08] * mterry high fives robert_ancell [23:08] thanks === broder_ is now known as broder [23:30] TheMuso: sure! that would be great, thanks! [23:41] mdeslaur: np, I'll assign the bug to myself then.