| blr | wgrant: hmm, which modal had the multi-step process, I thought it was 'link a related branch' on +bug, but that doesn't seem to be it | 03:04 |
|---|---|---|
| blr | I must be going mad, I was looking at it yesterday... | 03:06 |
| wgrant | blr: +sharing is multi-step | 04:41 |
| wgrant | blr: BugTask:+index's branch picker is highly customised. | 04:41 |
| wgrant | Well, not highly. | 04:41 |
| wgrant | But quite. | 04:41 |
| blr | wgrant: ah thanks, yes it was sharing. | 04:46 |
| satmandu | Quick question: I need to rebuild a package on my ppa for armh. armh builds have already been approved. How do I get an uploaded package to be rebuilt for armh? | 13:48 |
| satmandu | it seems that ubuntu-build might be the correct tool to use for this, but I can't figure out the right syntax for working with ppa builds | 13:48 |
| cjwatson | satmandu: If armhf was turned on after you uploaded the package in question, you'll need to upload it again. | 13:48 |
| cjwatson | That is, bump the version in debian/changelog but make no other changes. | 13:49 |
| satmandu | that's what I thought. ok thanks | 13:49 |
| cjwatson | I could check if you told us which PPA. | 13:49 |
| cjwatson | Oh, https://launchpad.net/~satadru-umich/+archive/ubuntu/sernet-samba/+packages I guess | 13:49 |
| satmandu | ppa:satadru-umich/sernet-samba | 13:50 |
| satmandu | yup that's it | 13:50 |
| cjwatson | Yes, bump version and reupload. Future uploads will have armhf builds created by default. | 13:50 |
| satmandu | ok will do. Thanks for the help! | 13:51 |
| === xnox_ is now known as xnox | ||
| satmandu | Can I ask a build question? | 16:08 |
| satmandu | my amd64 build took 16 min | 16:09 |
| satmandu | but the armhf build seems to be stuck. | 16:09 |
| satmandu | I want to be a good ubuntu citizen... | 16:09 |
| satmandu | is there anything I'm doing wrong? | 16:09 |
| satmandu | amd64 build: https://launchpad.net/~satadru-umich/+archive/ubuntu/sernet-samba/+build/7672673 | 16:10 |
| satmandu | current armh build: https://launchpad.net/builders/lgw01-53 | 16:10 |
| cjwatson | satmandu: armhf builds run via qemu-user-static. It's not very robust, and some things (particularly, but not limited to, just about anything involving threading) just plain don't work. | 17:03 |
| satmandu | Hmmm. ok. But the only way to store builds as a ppa is to have them compiled through the qemu-user-static, right? I can't build the packages locally and then upload them? | 17:04 |
| cjwatson | Correct. | 17:05 |
| satmandu | ok :) | 17:05 |
| cjwatson | We'll be bringing proper virtualised ARM builds online in the next couple of months or so. | 17:05 |
| satmandu | understood. | 17:05 |
| cjwatson | That should make many things better. | 17:06 |
| satmandu | Does the i386 build process often cause permissions issues during build? Getting a crash here: https://launchpadlibrarian.net/212115290/buildlog_ubuntu-vivid-i386.samba_99%3A4.1.19-11_BUILDING.txt.gz | 17:38 |
| cjwatson | satmandu: The difference between amd64 and i386 here is that (for >= vivid) amd64 builds the architecture-independent components as well. You can reproduce this locally using sbuild -A vs. sbuild --no-arch-all. | 17:41 |
| cjwatson | Also, that isn't a permissions issue; read the error message closely. | 17:41 |
| satmandu | the directory isn't being created... | 17:41 |
| cjwatson | Right, that's not permissions. :-) | 17:42 |
| satmandu | I think it's confused about wanting a /tmp/ folder | 17:42 |
| satmandu | but it works ok on amd64 | 17:42 |
| cjwatson | Your build is probably only creating it in the architecture-independent case. | 17:42 |
| satmandu | right | 17:42 |
| cjwatson | I don't see how /tmp/ has anything to do with it. | 17:43 |
| satmandu | just say /var/lib folders earlier as /tmp/var/lib etc etc... | 17:43 |
| satmandu | I'm just going to try these builds locally then | 17:43 |
| cjwatson | You mean debian/tmp/var/lib | 17:43 |
| satmandu | to see if they work | 17:43 |
| satmandu | yes | 17:44 |
| satmandu | Is there any way to see why this is stuck before I cancel it? https://launchpad.net/~satadru-umich/+archive/ubuntu/sernet-samba/+build/7673738 | 17:44 |
| cjwatson | satmandu: Not really, no. | 17:45 |
| cjwatson | Pretty sure your builds are just doomed on armhf for now :-( | 17:46 |
| satmandu | so it would seem | 17:46 |
| cjwatson | satmandu: sernet-samba-common is Architecture: all - this means that it is only built in builds that build architecture-independent packages. | 17:46 |
| cjwatson | satmandu: So debian/sernet-samba-common/<anything> won't have been created by other bits of the packaging process for architecture-dependent-only builds, such as i386. | 17:47 |
| cjwatson | satmandu: The correct fix with modern-ish versions of debhelper, such as that in vivid, is to change "override_dh_fixperms" to "override_dh_fixperms-indep", since you only need to override the arch-indep behaviour there. | 17:48 |
| satmandu | noted | 17:48 |
| satmandu | Is there a good way to build debs locally from the dsc & changes files to replicate what launchpad is doing? | 17:55 |
| satmandu | good = best practices | 17:55 |
| satmandu | aside from this: https://help.ubuntu.com/community/UpdatingADeb | 17:55 |
| cjwatson | satmandu: https://wiki.ubuntu.com/SimpleSbuild | 17:56 |
| blr | morning | 20:36 |
| cjwatson | evening | 20:52 |
| blr | how's things cjwatson? | 20:53 |
| cjwatson | I think I have most of the initial pieces for snap building put together, just got sidetracked into fixing https://bugs.launchpad.net/launchpad/+bug/1424672 | 20:53 |
| mup | Bug #1424672: LiveFS builds cancelled before they start sort above other builds in history <confusing-ui> <soyuz-build> <Launchpad itself:In Progress by cjwatson> <https://launchpad.net/bugs/1424672> | 20:53 |
| cjwatson | (since the livefs and snap code are *cough* suspiciously similar, so I didn't want to import that bug) | 20:53 |
| blr | cjwatson: hoping to get onto the squid authentication service soon, trying to get the 2-stage clientside ref picker sorted. | 20:54 |
| * cjwatson nods | 20:54 | |
| blr | discussed oauth/bearer token auth vs basic auth with wgrant yesterday, he seemed to think basic should suffice | 20:54 |
| cjwatson | I didn't totally follow, but the arguments sounded plausible :) | 21:02 |
| beuno | internal auth? we use basic auth everywhere I think, except for the things where it's easier to use tokens because they already terminate sso | 21:03 |
| satmandu | cjwatson: I'm trying one last armhf build.. which appears to be going albeit slowly. (The compile is running much faster on my raspberry pi sitting next to me.) | 21:07 |
| satmandu | So slow that I should cancel it to not be a drain on the build servers? | 21:07 |
| satmandu | https://launchpad.net/~satadru-umich/+archive/ubuntu/sernet-samba/+build/7674252 | 21:07 |
| cjwatson | satmandu: We have plenty of build resources at the moment. If it actually works then it's fine. | 21:22 |
| satmandu | cjwatson: Thanks. | 22:25 |
| satmandu | cjwatson: Does this look like a hung compile to you? https://launchpad.net/builders/lgw01-54 | 22:51 |
| satmandu | I can't debug it any further from my end. | 22:52 |
| wgrant | satmandu: qemu-user-static is quite slow. You pretty much have to watch the build log to see if it changes. | 22:54 |
| cjwatson | satmandu: Yeah, basically no way to tell. Leave it for a few hours; if it's idle with no output for too long then the builder should kill it anyway. | 23:01 |
| satmandu | Understood. Thanks again. | 23:14 |
| === Peng_ is now known as Peng | ||
| carstenh | Hi, loggerhead fails to display tabs as tabs in a diff. Tabs displayed as tabs: http://paste.ubuntu.com/11908410/ - tabs displayed as spaces: http://bazaar.launchpad.net/~ubuntu-core-dev/command-not-found/ubuntu/revision/190/zsh_command_not_found | 23:42 |
| carstenh | In case you ever wondered why cron logs three useless spaces every hour to syslog ... someone mixed up tabs with spaces in /etc/crontab, so this is not only an only academicially relevant issue | 23:42 |
| carstenh | And your paste service is suboptimal, correctly pasting a tab-idented file from a terminal into a browser is complicated and error-prone (at least I failed in doing so, I inserted an additional space). A way to upload a file and a way to download w/o logging in would be helpful (paste.debian.net is way better at this), it's also ugly in lynx (but paste.debian.net is even more ugly in lynx). Good night | 23:48 |
| cjwatson | pastebinit supports non-browser upload to paste.u.c | 23:49 |
| cjwatson | the login requirement is annoying but it mitigated some quite serious attacks aiui (I don't remember the details - it isn't a Launchpad service) | 23:50 |
| carstenh | Sounds good, such a tool is a great addition, but no replacement for an upload button in the html page. | 23:50 |
| cjwatson | we don't maintain it | 23:50 |
| cjwatson | loggerhead should be easily enough fixable, but could you please file a bug so we don't forget? | 23:51 |
| carstenh | Ok, I think the tab vs spaces issue in loggerhead is still valid | 23:51 |
| cjwatson | irc doesn't make a good bug tracker | 23:51 |
| carstenh | I'm goging to bed soonish, you could just paste what I wrote into a bug report since you are more familiar with launchpad :) | 23:52 |
| cjwatson | I'm going to bed *now*, and on phone so no easy paste | 23:52 |
| cjwatson | so no, not really ... | 23:53 |
| carstenh | Ok, I'll try to remember this when I login into launchpad next time :) | 23:53 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!