15:00 #startmeeting 15:00 Meeting started Wed Jul 31 15:00:37 2013 UTC. The chair is ev. Information about MeetBot at http://wiki.ubuntu.com/meetingology. 15:00 15:00 Available commands: #accept #accepted #action #agree #agreed #chair #commands #endmeeting #endvote #halp #help #idea #info #link #lurk #meetingname #meetingtopic #nick #progress #rejected #replay #restrictlogs #save #startmeeting #subtopic #topic #unchair #undo #unlurk #vote #voters #votesrequired 15:00 [TOPIC] Lightning round 15:00 $ echo $(shuf -e barry doko stgraber jodh ev cjwatson xnox) 15:01 stgraber xnox cjwatson doko ev barry jodh 15:01 slangasek not here? 15:01 He's at the Isle of Man client sprint and unavailable 15:02 xnox: I'm on vacation, so it's your turn 15:02 oh right, apols :) 15:02 * Last week wed-sun was on holiday 15:02 * From the week before, android and blobs are packaged and uploaded 15:02 into ubuntu universe. 15:02 * M.I.R. for above is filed, and in-progress (need to answer security 15:02 enquiries). 15:02 * TODO: automate "binNMU"/daily building of android package (such that 15:02 I'm not the only one who can/knows how to upload it) 15:02 * Working on sorting out btrfs mounting/fsck once and for all & get 15:02 btrfs in sync between Debian & Ubuntu. Should be fine with -5 15:02 update. 15:02 * Working on sorting out ubiquity: outstanding merge proposals & U1 15:02 xnox: Ah, you sorted out your internet troubles then 15:03 login page. 15:03 * Verified pending-removal SRU bugs for mdadm & partman-lvm 15:03 * Trying to help resolve bug 1197766, initially provided workaround 15:03 was not sufficient 15:03 bug 1197766 in OEM Priority Project "Different partition layout after recovery with keep home partition" [Critical,Confirmed] https://launchpad.net/bugs/1197766 15:03 .. 15:03 stgraber: touche =) i was out last week ;-) 15:03 cjwatson: yes. they trigger full line resynchronisation via automated telephone option selection. Which got my line back up. 15:04 so was only a short down-time <<30minuts. 15:04 Release engineering sprint. Spent most of the second half of the week implementing build cancellation: 15:04 - https://code.launchpad.net/~cjwatson/launchpad/buildstatus-aborted/+merge/176990 15:04 - https://code.launchpad.net/~cjwatson/launchpad-buildd/fix-abort/+merge/177003 15:04 - https://code.launchpad.net/~cjwatson/launchpad/buildmaster-cancel-properly/+merge/177580 15:05 I'm still awaiting some reviews and will probably need to make some changes, but the bulk of it is done. 15:05 foundations-1305-click-package: 15:05 - Various minor tweaks to prepare for the IoM demo. Bug 1204523 was especially entertaining to fix. 15:05 bug 1204523 in click (Ubuntu) "Click packages can't be installed from directories not viewable by the clickpkg user" [High,Fix released] https://launchpad.net/bugs/1204523 15:05 - Fixed rather embarrassing Unicode support failures. 15:05 foundations-1305-arm64-bringup: 15:05 - Bent juju-core to my will to bootstrap a new build farm, after the Canonistack LCY02 instance-deletion instance. 15:05 - Tried to get some saucy builds working. Currently blocked on some tail-ends of toolchain rebootstrapping, but should be able to start helping out again soon. 15:05 .. 15:05 - arm64 bootstrap (later more) 15:05 - MIR's 15:05 - binutils update 15:05 (done) 15:06 - I've spent much of the week continuing to get mobile error reporting up and 15:06 running. We now have whoopsie and apport on the Touch images as well as a 15:06 settings page for controlling reporting. I have a meeting later this week 15:06 with Katherine to discuss enabling reporting by default. 15:06 - I've run into what looks like a race condition on whoopsie startup on the 15:06 Touch images, which prevent it from uploading reports. I'm taking the 15:06 opportunity to build some more tests into things like whoopsie-preferences 15:06 (hooray for dbus-test-runner, by the way) as I get to the bottom of this. 15:06 - Work continues on the Cassandra crisis. You may have noticed that 15:06 https://errors.ubuntu.com is only showing data from the past week or so. We 15:06 had to cut over to a new database. Nothing is lost. We're backing up the old 15:06 database and will merge in the data when that is complete. If you don't 15:06 UGH. Die pidgin. 15:06 right, https://pastebin.canonical.com/95283/ 15:06 done! 15:07 ev: i don't use dbus-test-runner, but i managed to finally track down a very interesting (and annoying) behavior in the underlying dbus libraries ;) 15:07 errr http://paste.ubuntu.com/5933046/ :) 15:07 image based updates: LP: #1206515; LP: #1206558; LP: #1206523; LP: #1204528; LP: #1204976; various dbus api changes for better u/i integration; upload 0.9, 0.9.1, 0.9.2 15:07 barry: oh? 15:07 Launchpad bug 1206515 in Ubuntu system image "dbus service requires /tmp/system-image directory" [High,Fix committed] https://launchpad.net/bugs/1206515 15:07 Launchpad bug 1206558 in Ubuntu system image "system-image-dbus should run on system bus" [Critical,Fix released] https://launchpad.net/bugs/1206558 15:07 Launchpad bug 1206523 in Ubuntu system image "Need a system dbus policy for the system dbus service" [High,Fix released] https://launchpad.net/bugs/1206523 15:07 Launchpad bug 1204528 in Ubuntu system image "Add DBus mocks for u/i testing" [High,Fix released] https://launchpad.net/bugs/1204528 15:07 other: LP: #1073919; LP: #1205505; LP: #1181814; python-webob 1.2.3-3 (py3 support) and 1.2.3-4, debian bug 717655; PEP 394 (shebang lines); emacs 24.3 bug fix and upload; python-coverage review 15:07 Launchpad bug 1204976 in Ubuntu system image "Rename IsUpdateAvailable() and make it non-blocking" [High,Fix released] https://launchpad.net/bugs/1204976 15:07 ah nice 15:07 in progress: LP: #1206866, system-image 0.9.3 15:07 Launchpad bug 1073919 in command-not-found (Ubuntu Raring) "Crash with Python 3" [High,In progress] https://launchpad.net/bugs/1073919 15:07 Launchpad bug 1205505 in python-gnupg (Ubuntu) "Sync python-gnupg 0.3.4-1 (universe) from Debian unstable (main)" [Undecided,Fix released] https://launchpad.net/bugs/1205505 15:07 Launchpad bug 1181814 in wsjt (Ubuntu) "Crash: AttributeError: 'module' object has no attribute 'isNumberType'" [High,Confirmed] https://launchpad.net/bugs/1181814 15:07 Debian bug 717655 in python-webob "python-webob: Add python 3 support" [Wishlist,Fixed] http://bugs.debian.org/717655 15:07 Launchpad bug 1206866 in Ubuntu system image "Dies with dotted version numbers? - TypeError: unorderable types: Image() < Image()" [High,In progress] https://launchpad.net/bugs/1206866 15:08 ev: yeah, so i use $DBUS_SYSTEM_BUS_ADDRESS envar to rendezvous on a private bus... 15:08 turns out that once libdbus is intiialized, it never consults that envar again 15:08 How did the image-based-updates demo go? 15:08 ha! 15:09 so you can't change it and expect the same process to rendezvous on a different bus address 15:09 cjwatson: mixed i think. they got some weird bug that i don't understand at all but eventually got it to update and reboot 15:09 cjwatson: and of course it wfm here locally :/ 15:09 All the best bugs are like that 15:10 cjwatson: anyway, that's LP: #1206866, ongoing 15:10 Launchpad bug 1206866 in Ubuntu system image "Dies with dotted version numbers? - TypeError: unorderable types: Image() < Image()" [High,In progress] https://launchpad.net/bugs/1206866 15:10 * barry is very happy that phablet-flash now allows --ubuntu-bootstrap and -r -2 15:10 (Oh, I forgot to mention, I'm also working on implementing the "devel" series alias for saucy) 15:10 (allows you to flash to an older image) 15:10 cjwatson: nice 15:10 anyway... 15:10 done 15:10 :) 15:10 [TOPIC] Bugs 15:10 No Brian. 15:10 ev: paste.canonical.com instead of paste.ubuntu.com ? not using pastebinit? =) 15:11 Does anyone have bugs worth a mention? 15:11 xnox: that would've been the smart thing to do 15:11 erm, shall I go at the end? :) 15:11 OH HAI 15:11 jodh: you're up 15:11 Steve asked me to get back to bug 1065281, so I'm going to have to roll up my sleeves again on that :-/ 15:11 bug 1065281 in OEM Priority Project quantal "Installer crashed when trying to partition 4k/4k sector hard disks" [High,In progress] https://launchpad.net/bugs/1065281 15:11 soz. I copy and pasted from when you weren't here :) 15:11 But nothing interesting to talk about as yet 15:11 * foundations-1305-upstart-work-items: 15:11 - Finished the upstart-dconf-bridge and raised MP: 15:11 https://code.launchpad.net/~jamesodhunt/upstart/upstart-dconf-bridge/+merge/177650 15:11 - Upstart integration tests (DEP-8): 15:11 * ev nods 15:11 - Updated python Upstart test module and resubmitted MP: 15:11 https://code.launchpad.net/~jamesodhunt/upstart/python-upstart-module/+merge/157549 15:11 - Spent remainder of time working on DEP-8 scripts to make use of 15:11 nested pristine VM which will be provided by autopkgtest 15:11 environment. Attempted to use NBD to mount VM disk and configure 15:11 in chroot but requires too many hacks. Plan is now to boot VM, ssh 15:11 in to configure it, reboot, then run tests, but use NBD to collect 15:12 results data (ongoing). 15:12 * upstart: 15:12 - upstart-text-bridge simplified (local sockets only), renamed to 15:12 upstart-local-bridge and MP raised: 15:12 https://code.launchpad.net/~jamesodhunt/upstart/upstart-local-bridge/+merge/177027 15:12 - Investigated bugs 1205443 and 1202287 and merged the branch below that 15:12 Error: Launchpad bug 1205443 could not be found 15:12 Error: Launchpad bug 1202287 could not be found 15:12 doko: mostly not here, no 15:12 will hopefully resolve the issue: 15:12 lp:~jamesodhunt/upstart/quiesce-cleanup-fix 15:12 * misc 15:12 - debconf/LPC presentation: ongoing updates. 15:12 Ử 15:12 [TOPIC] arm64 bootstrap 15:12 ev: i finally got fixed a looooongggg standing emacs bug that was annoying the crap out of me for years :) 15:12 Steve asked doko to bring us up to speed on the arm64 bootstrap 15:12 With that, I yield to doko :) 15:13 barry: :D 15:13 Aarch64 bootstrap - how we did start ... 15:13 was the fix to use Vim instead? 15:13 * barry is not amused 15:13 instead ?? 15:13 (just waiting between my pastes ...) 15:13 - Start with armhf. It's in the archive, we know that it does build (or not). 15:13 - Build an armhf buildd chroot entirely with cross-built packages 15:13 - Use staged builds (DEB_STAGE=stage1) to break cyclic dependencies, or not to build unnecessary/unbuildable binary packages. 15:13 - Make sure that the cross-built chroot is usable. 15:13 the fix always involves vim ... no matter what fix :P 15:14 - Build and package the Aarch64 cross toolchain 15:14 - Start the cross build again, targeting Aarch64 15:14 - Fix Aarch64 specific bugs during that cross-build 15:14 - Finally had a build chroot + debhelper available (April) 15:14 - Make sure that it works using the foundations model (simulator). 15:14 - Building gnu-hello takes over an hour. 15:14 we got to this state in April 15:15 - During saucy ... start building native packages (with nocheck enabled). 15:15 - Build with nocheck. Good enogh if the package builds and can be used as a b-d 15:15 - We don't have any sbuild or dep-wait information, so set a goal to build a package. Starting with ... you guess it ... the toolchain. 15:15 - For warming up, binutils 15:15 Turns out we've built the toolchain several times ;-) 15:15 I think of it as a stress-test of "Reflections on Trusting Trust" 15:15 - Build-depends: bison, flex, gettext, texinfo, dejagnu, quilt, file, xz-utils, lsb-release, zlib1g-dev 15:15 - Identify the b-d's which are only needed for running the tests. Here: dejagnu 15:15 cjwatson, not just the toolchain =) 15:16 if gnu-hello takes over an hour, gcc 3 stage bootstrap took... weeks?! 15:16 - Next eglibc .. b-d: libaudit-dev 15:16 (http://cm.bell-labs.com/who/ken/trust.html in the unlikely event anyone hasn't encountered it) 15:16 ... audit: b-d: libkrb5-dev, libldap2-dev, libprelude-dev, and finally systemd 15:16 crap, maybe we come back later ... 15:16 - Next gcc ... easy as binutils 15:16 ... wait, except for gcj and needing gtk+2.0 ... 15:17 - There are a lot of cyclic dependencies in the gtk stack. 15:17 - Often not clear which ones are only used for testing. 15:17 - gtk b-d's on cups, so disable 15:17 - disable udeb builds, halfs the build time 15:17 - Ok, now gcc did build 15:17 - Next target: build LAMP ... 15:18 Break it down: 15:18 - build reduced build tools (cmake, doxygen without Qt) 15:18 - build reduced openldap, krb5, heimdal 15:18 - build reduced avahi 15:18 - build databases postgres and mysql (without krb, heimdal) 15:18 - build krb5, heimdal 15:18 - build cyrus-sasl2 15:18 - build apache 15:18 The object of the exercise of course is to get to the point where we can start normal-ish autobuilds as quickly as possible. Did I hear correctly that we are at the point of being able to self-host sbuild soon? 15:18 - build php5 15:18 - Any volunteers to test if that actually works? 15:19 Hm. I'm slightly surprised we're doing LAMP in stage1. I thought that was a stage2 target, or even something we'd want to push for early once we have builds in Launchpad 15:19 well, busted1.53 did fail to build twice, and took the simulator down 15:19 Of course the bits that involve circular build-deps are worth doing in stage1 15:19 * xnox giggles at busted1.53 15:20 it's systemd, audit and things like that needing the crypto stuff, and the crypto stuff the database stuff 15:20 so if sbuild can use dchroot, fine, else we need schroot first 15:20 - A lot of packages have b-d's for documentation builds not in build-indep ... these should be fixed. 15:20 - Finally I had to build texlive-bin. Running dh_autoreconf on 60+ configuries takes more than three days ... I start hating it. 15:20 I don't believe it can 15:21 dchroot is fairly dead except for the version built out of schroot 15:21 have to finish my other builds before I give it a try. but it's the last package 15:21 - And now we can build eglibc ... 15:21 - Not the only cycles. In a lot of cases you have to see where it is the easiest way to break a cycle. Don't touch "uncommon" build systems. 15:21 - Next steps: Get build-deps for qt4-x11 ready (almost done) 15:21 - gstreamer (avoid gsl), gst-plugins (avoid external plugins) 15:22 - Have a look at gtk+3.0 (install failure, don't understand it) 15:22 - Have a look at the sound stack (pulseaudio, alsa, bluez) 15:22 - Have a look at the video stack 15:22 - Have a look at guile. The built interpreter doesn't terminate 15:22 - Ocaml not yet ported, looks like other distros are working on that 15:22 - How to build initramfs (klibc isn't yet ported, only used by Debian/Ubuntu) 15:22 - Have a look at https://bugs.launchpad.net/linaro-aarch64 15:22 Help: How do identify both open and closed bugs, which either 15:22 don't have a distro task, or have an open distro task? 15:22 - Get dep-wait information for packages without trying to build them 15:22 Just takes too long ... 15:22 - Setup a sbuild which rebuilds already built but out-of-date packages 15:22 doko: Looked at edos-distcheck? 15:22 doko: gtk+3.0> what fails to install? 15:22 That's what Debian's wanna-build uses 15:22 not yet, chdist didn't work 15:23 chdist should work given that all sbuild is doing is an apt-get invocation 15:23 slangasek, this is strange, I'll post the build log once it finishes again 15:23 And all chdist does is set up apt to look at different sources 15:24 doko: i'd also like to see gtk+3.0 fail to install errors / log. 15:24 right, but it couldn't handle the saucy archive with the [armhf] identifier 15:24 so the other blocker is guile (and we need it for autogen) 15:24 - About packaging ... 15:24 - Unneeded build-deps, or ones which should be in build-indep 15:24 - shell script like makefiles (configure & build in one target) 15:24 - no support for parallel builds 15:24 - no support for staged builds 15:24 Still odd, since it doesn't parse sources.list itself :) 15:25 You might need to set APT::Architecture and/or APT::Architectures in the chdist environment's apt.conf 15:25 I can give it a try later, I didn't care that much at the moment 15:25 - About using the simulator: 15:25 - Crashes sometimes, having to restart builds 15:25 - The time is slower in the simulator. Uploading a package to saucy, then trying to build it has interesting effects (generated files are older than the one in the package). 15:26 would love to install ntp, but that b-d's on autogen ... 15:26 I was about to say :) 15:27 so, that's it what I did prepare ... 15:27 thanks doko 15:27 doko: how much is still being cross-built right now, vs. built under emulation? 15:28 slangasek, for stage1 now nothing. but we didn't re-build cross-built packages from stage0 yet 15:28 just when I noted issues, like the cross-built bash missing job control 15:28 * slangasek nods 15:28 [TOPIC] AOB 15:28 anything else? 15:29 We're keeping things reasonably separate between stage0 and stage1, but I'm working on a saucy-stage0 at the moment to see if I can make some cycles a bit easier to break 15:29 brief update from the sprint? :) 15:29 (Since the raring-stage0 generally didn't use its own output to build, so it's only the first layer in a lot of cases) 15:29 go for it :) 15:29 slangasek: please! 15:29 so click packages and system updates were both demoed this week 15:30 the click package demo went off without a hitch (beuno clearly spent more time rehearsing than I did) 15:30 very well-received 15:30 and very close to being ready to go live, which is exciting 15:31 the system updates demo had a few problems, but everything did come together well all things considered 15:31 beuno was doing little else but rehearsing from about last Wednesday :) 15:31 and I've done several real upgrades through the UI, the code really does work :) 15:31 so kudos to everyone who was involved in those two big pieces of work 15:32 well done, guys 15:32 everyone here is also happy to have whoopsie landed on the images 15:32 though I've just noticed that with the system images, /var/crash is read-only, so we have some more fixing to do there ;) 15:32 ha! 15:32 Zarro boogs found. 15:32 but the unity team are eagerly awaiting having their crashes recorded 15:33 That's good to know. 15:33 and there's pent-up demand for both the upstart android bridge and package cross-build support (a few more packages needed to support before it meets the team's needs) 15:33 so they're chomping at the bit 15:34 in short: you guys are all doing great work, which is greatly appreciated by the rest of the team 15:34 so keep it up :) 15:34 15:35 slangasek: if you see someone responsible for the current touch experience, spawn an app and then SIGSTOP it for me 15:35 then tell them to use the launcher 15:35 I've sent a mail to phablet about it 15:37 ev: this demos a bug I suppose, with the launcher locking up? 15:37 slangasek: if an application hangs, there is no way out 15:37 all input seems to be channelled through the foreground application 15:37 #endmeeting