=== juliank_ is now known as juliank
Bluefoxicywait, what03:21
Bluefoxicythere are no text terminals03:21
Bluefoxicyctrl+alt+F1-F6 just shows the line "Scanning for Btrfs filesystems"03:21
=== salem_ is now known as _salem
pittiGood morning05:51
tsimonq2o/ pitti, how are you? :)05:52
pittihey tsimonq2! a bit tired, but ok, how about you?05:55
tsimonq2great pitti :)05:59
tsimonq2pitti: quick question, are you a Core Dev?06:00
tsimonq2pitti: you on the release team?06:01
pittihttps://launchpad.net/~pitti/+participation :)06:01
tsimonq2oh sorry06:02
tsimonq2yeah nevermind06:03
pittididrocks: bonjour !06:45
pittididrocks: do you know about the ubuntu-make test regressions? (http://autopkgtest.ubuntu.com/packages/u/ubuntu-make/yakkety/amd64/) -- it seems to complain about "debian/tests/collect-coverage: line 8: cd: /tmp/global-coverage: No such file or directory"06:46
pittioh, and "Couldn't find static file 'jquery.ba-throttle-debounce.min.js'" in the "small" test06:46
pittithe one above might just be a consequence of that06:47
didrockshey pitti! this error happen when no tests have been running.06:48
didrocksmaybe one build-dep breaking the coverage report?06:48
pittididrocks: right, so I guess the missing jquery.ba-throttle-debounce.min.js causes the failure?06:48
didrocksprobably yeah (same, this is on the coverage report side)06:49
cpaelzerlate but good morning everybody06:50
pittijuliank: hm, didn't we use to have something like $APTROOT?08:48
pittijuliank: i. e. if I have my little sandbox with foo/etc/apt/sources.list and foo/var/lib/apt/lists etc.08:49
pittijuliank: (working on the archive auth -- I'd rather install the GPG key properly than [trusted=yes]08:49
pittithe apport code uses apt.Cache(rootdir=), but that still barfs on the signature error even though trusted.gpg.d/ddebs.ubuntu.com.gpg is in place08:50
pittioh, I think apt -o RootDir="$(pwd)" update08:52
pittijuliank: meh, I can't get this to work :( <myroot>/etc/apt/trusted.gpg.d seems to get entirely ignored, it always takes the one in the root dir09:20
pittijuliank: so at least test_install_package_from_a_ppa() is a regression in apt, the key for the PPA is definitively there; the test_install_old_packages() failure is my fault, I'll push the fix (but that's not sufficient for the same reason)09:28
Saviqpitti, morning, can you please rerun https://requests.ci-train.ubuntu.com/static/britney/ticket-1604/landing-019-yakkety/excuses.html with proposed enabled, it's fallout from the Qt 5.6.1 release09:35
Saviq(the in progress ones will fail, too)09:35
pittijuliank: filed as bug 1607283 with a standalone simple reproducer09:41
ubottubug 1607283 in apt (Ubuntu) "1.3~pre2ubuntu2 regression: trusted.gpg.d not read any more from apt root" [Undecided,New] https://launchpad.net/bugs/160728309:41
juliankpitti: Reproduced.09:44
pittijuliank: it could of course be a red herring and reading trusted.gpg.d from the sandbox may have never worked -- but apt previously didn't complain about it?09:45
juliankpitti: I'm going to bisect this09:45
pittijuliank: cheers!09:45
pittiSaviq: retried the failures and running ones09:45
Saviqpitti, thanks!09:46
juliankpitti: Can even reproduce it with 1.2.10 :/ - so it seems that's not a regression09:56
pittijuliank: hm, it works on my laptop with current yakkety10:03
pittiand apport with 1.3~exp1 doesn't fail on that10:03
juliankpitti: Well, failing is a strong word... 1.3~exp1 shouldn't find the keys, but only warns about the missing keys, and does not produce an error10:07
juliankpitti: https://paste.debian.net/785744/10:08
juliankpitti: Probably because apt-key can only look at config files, and not config values set in the parent process.10:10
juliankSo, you'd have to specify dir in a apt.conf file and then set APT_CONFIG env var to that path10:12
juliankNot sure how to fix that yet10:13
juliank We should probably dump the config to a file, set APT_CONFIG=temp-file and only then start apt-key10:14
pittijuliank: so apt.Cache(rootdir=foo) sets "Dir=foo" more or less?10:27
juliankIt sets Dir in the current process, but that's not forwarded to the apt-key process10:28
pittijuliank: oh, you mean apt-get update calls apt-key for verifying the sig, that's not done in-process?10:28
juliankapt calls apt-key which calls gpgv10:29
pittijuliank: so how did that work in earlier versions, did it just silently ignore the missing key?10:29
juliankIt should not ignore it silently10:29
pittion my laptop I don't have ddebs in "apt-key list" and yet that python3 command line works fine10:29
pitti(and the apport tests)10:29
Mirvpitti: Saviq: https://requests.ci-train.ubuntu.com/static/britney/ticket-1604/landing-019-yakkety/excuses.html just updated but I don't find any running or completed results that would have been run with yakkety-proposed. I do find results that were started eg 30mins ago10:43
SaviqMirv, pitti, yeah I was looking there too, I saw them initially with all-proposed in running queue, not sure if all the tests were restarted or just unity8... not sure where to look for recently-completed runs for PPAs10:44
pittino, I restarted all the regressions and all teh runing ones with --all-proposed10:45
pittiseems that didn't help indeed10:45
Saviqpitti, e.g. https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-yakkety-ci-train-ppa-service-landing-019/yakkety/amd64/u/unity8/20160728_094920@/log.gz still doesn't seem to have proposed enabled so maybe we're looking at the tests that were running already?10:46
Saviqthose http://autopkgtest.ubuntu.com/running.shtml#pkg-unity8 however don't have "All-proposed: True"10:47
Saviqfor ppa 01910:47
MirvSaviq: that started after your discussion, and there are similar ones 15-20mins later that also don't have proposed enabled10:47
pittiSaviq: landing-060 != landing-01910:48
Saviqpitti, there's some landing-019 further down10:48
pittiwell, I did queue them with all-proposed, apparently that didn't make it to the testbed for some reason10:48
juliankpitti: RFC: https://github.com/Debian/apt/compare/master...julian-klode:bugfix/apt-key-config10:48
juliankOr rather just https://github.com/Debian/apt/commit/1bdb1e50e2b36bd5ffd14b8dbec8be1052eb198710:49
pittijuliank: OOI, what does that new "conf" temp file do? I thought this  was just about setting $APT_CONFIG?10:50
juliankpitti: We dump the current in-process config to a temporary file, and then pass that as APT_CONFIG to apt-key10:50
juliankIt's ugly, but I don't really see what else we could do10:51
pittiSaviq, Mirv: I retried unity8/i386 again with --all-proposed (can do the others if that works)10:52
pittioh crap, I know10:52
pittiI did -s yakkety10:52
juliankI'm not sure if it is safe10:52
pittiwe want xenial I guess10:52
juliankWe mkstemp() the file, but then basically ignore the fd, and re-open by name for the C++ ostream10:52
pittino, actually, you do want y10:53
juliankI could use __gnu_cxx::stdio_filebuf() to directly use the fd10:54
pittiit's running now, I'll check in a bit10:54
juliankon the other hand, I also pass it by name to apt-key...10:54
juliankwe also pass the temporary split gpg signature + data to apt-key by-name, though...10:55
pitti$ APT_CONFIG=/tmp/sandbox python3 -c 'import apt; c=apt.Cache(rootdir="/tmp/sandbox"); c.update()'10:55
pittijuliank: ^ that also doesn't work though, should it?10:55
juliankpitti: No10:55
pittiI guess that's the part of "need to actually write the config file"10:55
juliankpitti: APT_CONFIG is a config file, not a dir10:55
juliankI guess I also do not set cloexec on the ofstream currently :/10:56
pittiok, so APT_CONFIG=/tmp/sandbox/etc/apt/apt.conf also doesn't work as I suppose that doesn't look for the other files relative to that10:56
juliankpitti: You'd need to have Dir "my-chroot-path" in your APT_CONFIG file10:57
pittiLaney: yay freeze lift10:57
pittijuliank: indeed, that does the trick; so that's what your branch does automatically now10:58
pittijuliank: nice, thanks muchly for fixing that!10:58
pittijuliank: I guess I'll upload apport then so that the next apt upload catches a fixed one (for the missing ddebs.u.c. key import)10:58
=== hikiko is now known as hikiko|ln
juliankpitti: Yes, I'll upload a new apt soon too to see if that works :)10:59
juliankI'll also upload a python-apt that tells apt-key to shut up, so we do not get stderr failures for python-apt11:00
juliankOnce everything works here, I'll upload it to Debian with possibly some more fixes from DonKult and sync it11:01
juliankhere = in Ubuntu11:01
julianknot on my machine :D11:01
pittijuliank: new p-apt just landed, FYI11:04
dokoMirv, did you have a chance to look at the powerpc related GCC6 qt*-opensource-src FTBFS?11:16
dokotvoss, sil2100: please could you fix the GCC6 related ftbfs for dbus-cpp and ubuntu-ui-tookit? http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20160705-gcc6-yakkety.html11:17
juliankpitti: uploaded python-apt (to fix the stderr output during test) and apt (with the apt-key config fix from my branch).11:31
* juliank still has to write a test case and probably refactor it a bit for upstream11:31
juliankFor now, the main point is to get the autopkgtests working again, though.11:32
Mirvdoko: not much, but note that yakkety-proposed now has Qt 5.6.1. looking at the previous rebuilds, all of them say "/usr/bin/ld.gold: --secure-plt: unknown option"11:33
Mirvso maybe it's the powerpc version of ld.gold11:33
=== _salem is now known as salem_
ubottullvm.org bug 20572 in Backend: PowerPC "PowerPC backend does not support secure PLT" [Normal,New]11:38
Mirvthat's LLVM though..11:38
dokoMirv, but you don't use clang explicitly, do you?11:39
Mirvdoko: no, just educating myself of the switch via google11:41
jgrimmdannf, would you be able to verify https://bugs.launchpad.net/ubuntu/+source/docker.io/+bug/157490411:47
ubottuLaunchpad bug 1574904 in docker.io (Ubuntu Xenial) "Old clients cannot talk to Docker in 16.04" [High,Fix committed]11:47
Mirvpitti: Saviq: no proposed in unity8/i386 https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-yakkety-ci-train-ppa-service-landing-019/yakkety/i386/u/unity8/20160728_105414@/log.gz11:51
=== hikiko|ln is now known as hikiko
dokotedg, could you have a look at https://launchpad.net/ubuntu/+archive/test-rebuild-20160705-gcc6/+build/10416146 (GCC 6 FTBFS)?11:55
dokoMirv, looks like qtbase-opensource-src has configure options/values -use-gold-linker/-no-use-gold-linker/USE_GOLD_LINKER/CFG_USE_GOLD_LINKER. please disable then on powerpc. gold on powerpc doesn't have this option12:02
dokobut wondering why you only see this with GCC 6 ...12:03
pittixnox: do you think the new mdadm is ready for merging now?12:32
xnoxpitti, yes. still need to retest on ubuntu. ideally i would like to just port the remaining ubuntu-only bits to debian (e.g. apport integration) such that we could have it in-sync.12:39
xnoxpitti, are you volunteering to merge mdadm? =)12:40
pittixnox: I'm asking because systemd 231-1 needs the new version; I could also temporarily drop that Breaks:, I don't think we actually have an rcS init.d script in Ubuntu; I asked because if you were going to merge/sync anyway that wouldn't be needed12:40
pittixnox: I can do the merge (apport bits etc.), but I would suck at testing it12:40
pittixnox: AFAIR our main delta was udev vs. init.d script12:41
xnoxoh that. ok. I see.12:41
pittinot sure if that's still relevant with your rework, as that now seems to be dynamic in the Debian version too?12:42
xnoxI'm hoping to do openssl, boost1.61 and mdadm this week =) today and tomorrow. I guess i could bump mdadm above boost1.61 =)12:42
pittiI'm just done with ebtables, now going to nfs-utils (that's a big one), and scsitools/ifscheme (these shuold be small)12:43
pittithen the only remaining one is mdadm12:43
pittiand then we finally got rid of rcS.d init scripts :)12:43
xnoxpitti, it's now much better in debian than ubuntu, cause it now integrates with block-local loop for proper mount handling.12:43
pittisounds yummy :)12:43
pittinice work!12:43
xnoxpitti, may i ask you something about postgresql? =)12:43
pittixnox: sure, you can always ask (I've been out of the loop for a fair while, so not sure I can answer)12:44
xnoxwhy the slice is not named as "postgresql-slice" =)12:44
xnoxwhy the slice is not named as "postgresql-system.slice"12:44
xnoxbecause at the moment: systemctl status postgresql* -> will show "master" unit & all instances, but not the slice which has e.g. current memory usage stats =(12:44
xnoximho it should be renamed "system-postgresql.slice" -> "postgresql-system.slice"12:45
* xnox off to get coffee12:45
pittislices don't work that way, you can't arbitrarily name them; their name reflects the cgroup hierarchy12:45
pittii. e. all system services are in, or in a child of system.slice12:46
pittipostgresql.slice would be completely independent, thus global resource limits would then not apply to it any more12:47
pittixnox: status '*postgresql*' should work?12:47
SaviqMirv, yeah I just saw, not sure what's going on... pitti any idea why your "all-proposed" reruns end up running without proposed again https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-yakkety-ci-train-ppa-service-landing-019/yakkety/i386/u/unity8/20160728_105414@/log.gz ?¿12:52
pittiSaviq: oh, I know: Mirv requested it! bug 154133412:55
ubottubug 1541334 in autopkgtest (Ubuntu) "Do not run silo tests against all of -proposed" [Undecided,Fix released] https://launchpad.net/bugs/154133412:55
pittithis was changed in https://git.launchpad.net/~ubuntu-release/+git/autopkgtest-cloud/commit/?id=b029faacbe12:55
* Saviq shakes fist12:55
pittii. e. we don't run silos against -proposed because that broke train landings in other ways12:55
pittican't have it both ways :/12:55
pittiso I guess Qt needs to land first?12:55
pittihow can the new Qt break that silo if it is only in -proposed? or does the new silo depend on the new qt already?12:56
Saviqpitti, it builds against proposed12:56
Saviqpitti, so it links against the new qt12:56
pittia silo cannot land if any of its dependencies are stuck in -proposed anyway?12:56
SaviqI think it can12:57
pittiso I could temporarily revert that, but it will again break the cases from teh bug above12:57
Saviqbut here the problem is not dependencies but broken test runs12:57
Saviqpitti, couldn't --all-proposed override that? i.e. when you explicitly ask for it, shouldn't it go for all proposed anyway?12:58
pittiI guess so, yes12:58
Saviqthat would be best IMO12:58
MirvSaviq: pitti: yes, having proposed enabled brings usually more trouble than it helps, that's why it's generally disabled. but in case of transition like that, it could be useful to have an option to run with all proposed (even risking the other trouble it brings)13:02
pitticode updated/deployed, retried unity8/i38613:03
pittiif that works, I'll retry the others13:04
Saviqgreat, thanks13:04
xnoxpitti, ahhhh i did not realise that it is `systemd-escape`d path =)13:15
xnoxso it's like .device/.mount units13:15
pittixnox: right13:15
Saviqpitti, looking at http://autopkgtest.ubuntu.com/running.shtml#pkg-unity8 it's much better, so if you could restart all regressions in https://requests.ci-train.ubuntu.com/static/britney/ticket-1604/landing-019-yakkety/excuses.html with allproposed - thanks!14:33
LocutusOfBorganybody knows why mdadm is not installable on yakkety? it tries to remove init14:41
LocutusOfBorgpitti, ^^ seems somewhere systemd/init related14:41
mterryLaney, any luck on unity-greeter (I'm guessing no)14:42
Laneymterry: hey there, nope - haven't tried any more this week really14:45
mterryugh.  I don't want to be the one holding up gtk3.2014:46
Laneymterry: I would probably not actually hold it up on this bug14:46
mterryLaney, heh14:46
mterryWell I don't want to be the one breaking user logins when they make a typo14:46
pittiSaviq: indeed, done14:47
Laneymterry: I noticed that if you go down to the guest session entry and then back then it appears again14:47
pittiLocutusOfBorg: that only affects -proposed, that's known; see discussion with xnox two hours ago14:47
mterryLaney, yeah there's an easy way to get it back14:47
mterryLaney, but you have to try that.  And have more than one entry14:47
pittiLocutusOfBorg: i. e. systemd will stay in -proposed until mdadm gets merged, or I do a temporary ubuntu delta to drop the breaks14:47
Laneymterry: damn, I aws hoping that would be a clue to you14:48
mterryLaney, no.  I don't get it at all14:48
LaneyI think that robert_ancell is back now14:48
Laneyat least he emailed the team aboutu some work stuff...14:48
mterryLaney, we do set css class names somewhere in there...  but I didn't think we did fancy stuff with it14:48
Saviqpitti, awesome, thanks - Mirv it's running well now14:48
xnoxpitti, i have a merge ready, and will upload it soon, once i test that it is sort of correct =)14:49
xnox(it will be a sync)14:49
LocutusOfBorgI see now14:50
LocutusOfBorgI'll wait then, a bunch of packages are FTBFS because of this issue14:50
LocutusOfBorge.g. diffoscope14:50
pittiurgh, b-dep on systemd and mdadm??14:50
LocutusOfBorg sbuild-build-depends-diffoscope-dummy : Depends: python3-guestfs but it is not going to be installed14:50
Laneymterry: I think all of that should work with the rejigged CSS that I did14:50
LocutusOfBorgpitti, not directly but meh14:51
LocutusOfBorglibguestfs should be the one needing them both?14:52
pittiright, that's the one that is uninstallable right now14:52
xnoxLocutusOfBorg, why do you have proposed enabled on a development machine?14:52
pittiLocutusOfBorg: sorry about that -- should be fixed soon14:52
LocutusOfBorgxnox, I don't have14:52
xnoxLocutusOfBorg, one must not run with -proposed enabled on a development series.14:53
LocutusOfBorgI don't even run yakkety14:53
pittixnox: it's an FTBFS (which runs against -proposed)14:53
LocutusOfBorgyes exactly14:53
LocutusOfBorgI was looking at gammaray, and I saw diffoscope FTBFS14:53
mterryYup commenting out all css use in unity-greeter doesn't fix anything14:59
Laneyprobably punt to Robert for now15:01
Laneyhe should know lightdm signals the greeter to re-display the box15:01
Laneythat's where I would start poking, but it is pretty complicated to me atm15:01
seb128didrocks, pitti, could you comment on bug #1584575 maybe? I think it makes sense but you are more familiar with the work done there I think15:04
ubottubug 1584575 in lightdm (Ubuntu) "/lib/systemd/system/lightdm.service file has no [Install] clause" [Medium,New] https://launchpad.net/bugs/158457515:04
seb128like should it have a15:04
seb128    [Install]15:04
seb128    WantedBy=multi-user.target15:04
seb128    Alias=display-manager.service15:05
didrocksI don't remember the latest on this, as this got discussed over and over again, but I thought we decided on the Alias15:05
didrockslet's wait on pitti for now, he may have a better memory that I do on this15:06
pittiit's blurry as well, but indeed this is controlled by display-manager.service15:06
mterryLaney, hrm.  Found another bug.  If you click on the session-chooser icon, nothing is drawn and you can't go back15:07
pittiIIRC all *dm's postinsts need to set /etc/systemd/system/display-manager.service to whatever was chosen in debconf15:07
pittiseb128: ^15:07
pittisetting Alias= sounds correct, but not WantedBy=15:08
pittiwe don't want *all* DMs to start15:08
seb128good point15:08
pittithe whole mechanics for that is still horribly complicated and redundant -- you can use update-rc.d, debconf, /etc/X11/default-display-manager etc.15:08
pittiI think /etc/X11/default-display-manager is still meant to be the canonical way to configure this, that's why we skipped the [Install] sections so that you don't introduce contradictions with systemctl enable/disable15:09
seb128pitti, ok, so looks like a non-trivial one, might just be better to not change anything for now then15:12
Laneymterry: meh15:15
Laneyit works if you go into the inspector and some rule15:15
Laneylike * { border: 1px solid red; }15:15
Laneythen even if you pause it, it works15:15
mterryLaney, wait how do you use the inspector again?  you said ctrl+alt+i, which doesn't do anything for me.  I assume I need to be running some inspector tool first?15:17
Laneymterry: apt install libgtk-3-dev; gsettings set org.gtk.Settings.Debug enable-inspector-keybinding true15:19
Laneyshould have said, sorry :(15:19
mterrystill doesn't do anything for me  :(15:24
mterryWell both the prompt box and the session list use a PromptBox class to do drawing.  And it certainly has some weird drawing logic...  But per my tests, everything seems to be right15:25
pittiMirv, Saviq: ah, first green on https://requests.ci-train.ubuntu.com/static/britney/ticket-1604/landing-019-yakkety/excuses.html :)15:27
juliankpitti: So, the python-apt upload is basically done with testing. No new regressions, as expected, but it will need some manual work due to the click issue (ubuntu-make failure is ignored, click failure is not ) - that fixes the regression with the new APT. I also noticed that the apport test had a communication issue with LP on amd64 and retried that15:33
juliankSo, I think apt should land tomorrow :)15:34
pittijuliank: indeed, python-apt is all good except click, I'll hint that15:34
pittijuliank: thanks for retrying apport; once that lands this ought to help with the new apt15:37
pittijuliank: just weird that apport in http://people.canonical.com/~ubuntu-archive/proposed-migration/update_excuses.html#apt still failed the "PPA" tests that should work with the apt fix15:39
juliankpitti: Hmm, not sure15:40
juliankBut if there's still something to fix, we can do so15:42
xnoxpitti, mdadm 3.4-4 uploaded -> i will try to sync it when it becomes available for syncage. Or if you spot it first, feel free to sync it over. with force.16:11
=== gshereme is now known as gshereme|away
pitti"Use the force, Luke^WLaunchpad!" :)17:56
Mirvpitti: somehow armhf failures show still not using yakkety-proposed https://requests.ci-train.ubuntu.com/static/britney/ticket-1604/landing-019-yakkety/excuses.html - I've pinged Savi_q regarding the unity8 qmluitests failures17:57
Mirv(on x86)17:57
pittiMirv: oh, they ran there -- forgot to update these boxes17:57
Mirvok :)17:58
pittiMirv: updated the lxd boxes, retried everything again17:58
* pitti waves good evening17:59
Mirvpitti: thank you, good evening!17:59
=== NCommander is now known as mcasadevall
Guest55167does this channel cover 16.10?20:49
Guest55167anyone here?20:52
=== JanC is now known as Guest85214
=== JanC_ is now known as JanC
xnoxIs this Richard Brandson (Virgin Group) putting on Ubuntu cap? https://res.cloudinary.com/www-virgin-com/f_auto,q_80/virgin-com-prod/sites/virgin.com/files/lewis_hat_2.jpg22:37
infinityxnox: http://ubuntuismclothing.com/22:41
xnoxOMG there is a whole clothing line.22:41
* xnox throws all my clothes in suitcase and takes it to the dumpster22:42
=== salem_ is now known as _salem

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