IRC logs for #baserock for Tuesday, 2015-04-07

*** petefoth_ has joined #baserock04:36
*** petefoth has quit IRC04:37
*** petefoth_ is now known as petefoth04:37
*** persia_ has quit IRC04:44
*** persia_ has joined #baserock04:45
*** petefoth has quit IRC05:05
*** mike has joined #baserock06:10
*** mike is now known as Guest5861906:11
*** petefoth has joined #baserock06:28
*** franred has joined #baserock07:16
*** Albert has joined #baserock07:40
*** jonathanmaw has joined #baserock07:51
*** mdunford has joined #baserock07:57
*** gary_perkins has joined #baserock07:57
*** tiagogomes has joined #baserock08:02
*** bashrc_ has joined #baserock08:04
*** rdale has joined #baserock08:06
*** ssam2 has joined #baserock08:49
*** ChanServ sets mode: +v ssam208:49
*** CTtpollard has joined #baserock08:51
*** Krin has joined #baserock08:52
richard_mawDoes anyone have time to look at some BSP patches?08:55
richard_mawI submitted the BSP stuff we needed for OpenStack as https://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/definitions+branch:master+topic:openstack-bsp08:57
*** edcragg has joined #baserock08:58
richard_mawdon't worry about the merge conflicts, it's from gerrit not handling patch series' particularly nicely08:58
richard_mawthe merge conflicts should go away if it's merged in dependency order08:58
* straycat thought it was due to IPV6 being after those patches were created08:59
straycat*being enabled08:59
* richard_maw has the ability to +2 and merge them through, but would much prefer extra eyes so there's other people aware of what we're up to08:59
* straycat assumes they've been built and tested09:01
richard_mawstraycat: multiple times09:01
richard_mawI thought I'd rebased after the IPv6 was merged, but looking at the commit times, they are close enough that it may not have happened09:02
richard_mawstraycat: nope, definitely not the cause, I'd rebased on top of the IPV6 change09:04
straycatokay i didn't fetch the changes from gerrit, so it's possible i checked out the wrong thing09:05
richard_mawyeah, the openstack-v5-kernel branch pushed to git.baserock.org isn't the rebased version09:06
straycatyes i just downloaded the change from gerrit and it is based on ipv6, so i guess that's not the problem09:07
straycatrichard_maw, okay +1'd all the things on the basis that you've already tested it and none of it looks insane, that's the best i can do really09:11
richard_mawedcragg's armv8 BSP patches https://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/definitions+branch:master+topic:baserock/edcragg/armv8b64-definitions and https://gerrit.baserock.org/#/q/status:open+project:baserock/baserock/definitions+branch:master+topic:baserock/tiago-ed/moonshot-deployment also need more eyes09:13
*** persia is now known as persia__09:14
*** persia_ is now known as persia09:14
richard_mawthough I don't think all of the moonshot deployment patches are applicable, since they depend on https://gerrit.baserock.org/#/c/102/09:14
*** persia__ is now known as persia_09:14
straycatfranred, thanks for merging my change :)09:15
franredstraycat, no probs :)09:17
ssam2richard_maw: i looked at them and was not sure whether to +1 or +2. Can +2 if you want (i don't have any reason that they shouldn't be merged)09:22
richard_mawssam2: that's fine, I can +2 and merge them myself since we've got the requisite 2 +1s09:26
*** zoli__ has joined #baserock09:33
persiacatching up on backscroll: qemu-user-static for armv5 on x86 is nearly entirely feature-complete, and I've previously found it a sensible way to do armv5 work (as finding fast armv5 is hard).09:56
persiaThe rule for what can be built on what is set by the linux personality features.  setarch is the relevant userspace configuration tool.  Not every subarch can do every other subarch, but the code in setarch is incomplete.09:57
persiaFor ARM, ARMv8 can do ARMv8 and ARMv7, but not earlier.09:57
persiaARMv7 can do ARMv6 and ARMv5, but not earlier09:57
persiaARMv6 can do at least ARMv5 and ARMv4: I don't know about earlier.09:58
*** fay_ has quit IRC10:06
gary_perkinsHi, I'm told by rdale that morph uses git:// style urls, not ssh://git@ urls. Is this by design?10:09
*** fay_ has joined #baserock10:09
persiaOh, and on qemu-user-static: if you set up your binfmts right, you can just unpack a blob and chroot() normally.10:10
richard_mawgary_perkins: morph uses both in different contexts10:10
persiagary_perkins: Yes.10:10
richard_mawwe prefer git:// urls for reads, but you need a ssh://git@ url for writes10:10
gary_perkinsok, thanks for the info. It's just that we havn't setup a route for trove.codethink to be accessible for VPN users yet. So there is just an SSH proxy in place at the moment10:11
straycatyou can also use the git@host:repo form, though morph makes some assumptions about urlparse which leads to some unexpected directory name: git@host, but it still works10:12
straycathttps://github.com/pypa/pip/pull/2371 :(10:22
jonathanmaw:/ pygtk requires pygobject-2.0, which is incompatible with baserock's glib. making it use pygobject-3.0 doesn't work, since it's expecting a 'codegen' program of some kind, which pygobject-3.0 doesn't seem to provide.10:27
franredI want to enable python api in kmod, this is required for openstack (more specific for iscsi software). For this kmod needs cython so cython has to be moved to even core.morph or foundation.morph. Actually is in python-core. does anyone see weird to move cython to core?10:30
jjardonfranred: is it a build or a runtime dependency?10:32
franredjjardon, it is a build dependency10:35
bashrc_does anyone know if it's possible to check out a topic branch from gerrit?10:43
straycatbashrc_, if you find the number of the change on the tip of the topic branch and then use git-review -d <change number> then you'll get the entire topic10:44
bashrc_ah ok, havn't used git-review yet10:44
*** gary_perkins has quit IRC10:46
straycatit does make it all a bit easier10:46
*** gary_perkins has joined #baserock10:47
straycati'm sure you can also do that without git-review but i don't know how10:47
jjardonjonathanmaw: in what sense pygobject-2.0 is incompatible with baserock glib? AFAIK, There was not any api breackage in glib10:49
persiaWithout git-review, one fetches the specific ref from gerrit, and then checks out FETCH_HEAD, which includes the parents, etc.10:52
persiaUsing git-review is much easier.10:52
ssam2mason is red, but I've cleaned up lots of disk space on cache.baserock.org so it should go green again11:02
pedroalvarezstraycat: some days late but: pong11:07
straycatpedroalvarez, hey :)11:13
*** flatmush has quit IRC11:14
*** flatmush has joined #baserock11:27
*** zoli__ has quit IRC11:29
*** zoli___ has joined #baserock11:29
ratmice__ssam2: agree that ldconfig error is harmless, its actually the one which doesn't get the error that concerns me, which would seem to mean either ldconfig isn't being run, or is scanning the wrong directories, or?11:30
*** zoli___ has quit IRC11:32
jjardonfranred: if its a kmod dependency, I think move it to core is fine11:32
franredjjardon, Im currently writing the patch for this :)11:33
jonathanmawjjardon: full build failure log: http://paste.baserock.org/uloxojacok a whole lot of stuff is apparently deprecated11:51
jonathanmawbut those are just warnings, the actual errors appear to be the "could not write function" lines.11:51
*** nowster has joined #baserock11:53
nowster"The site baserock.branchable.com is not here yet." Is that something I should be concerned about?11:54
*** zoli__ has joined #baserock11:55
pedroalvareznowster: is not loading for you?11:55
nowsternope11:55
nowsterThis is editing the wiki11:55
pedroalvarezah11:55
nowsterOpenID moans about Branchable's OpenID policy, too.11:56
ssam2ratmice__: I'm pretty sure the warning does happen, it just gets logged into a file rather than to the console11:56
jjardonjonathanmaw: what branch/tag are you trying to compile? pygobject-2-28 branch builds fine here (latest glib, but not a baserock system)11:59
nowsterHas the procedure for updating the wiki changed?12:17
ratmice__ssam2: ahh, i was assuming that was a part of a diff between logfiles12:17
ssam2ratmice__: ok! the situation is two different build tools, so it's probably just a case of the logfiles being arranged differently or something12:18
ssam2nowster: the procedure is 'do it', as far as I know12:18
nowsterssam2: in that case, the sign in may be broken12:18
nowsterI can neither log in to edit using either Google or OpenID.12:19
ssam2nowster: hmm. I get an error when using http://openid.baserock.org/12:19
ssam2that used to work, so I think Branchable have broken something12:20
ssam2yes, using my StackOverflow OpenID also fails12:20
ssam2I can file an issue with Branchable after lunch, if nobody beats me to it12:21
ssam2we don't host the wiki, so there's not much else I can do12:21
jonathanmawjjardon: I'm using tag PYGOBJECT_2_28_612:26
jonathanmawI'll try the branch to see if it's different.12:26
jonathanmawjjardon: that branch worked.12:34
jjardonjonathanmaw: \o/12:34
* petefoth is trying to make the words 'Disambiguate definitions' fit the tune a d rhythm of 'Accentuate the positive' https://www.youtube.com/watch?v=fZUmAbi0Vm413:04
*** Albert has quit IRC13:27
ssam2is anyone still having an issue logging in to baserock.branchable.com ? I find that in firefox I get the ' The site baserock.branchable.com is not here yet.' error, but in Epiphany I can log in fine13:28
ssam2so maybe the problem is fixed but my cache is stale13:28
pedroalvarezjust logged in without problem13:31
*** mdunford has quit IRC13:33
*** Albert has joined #baserock13:33
*** Albert has joined #baserock13:33
*** mdunford has joined #baserock13:47
*** mdunford has quit IRC13:54
*** mdunford has joined #baserock14:00
straycatoh cool all the openstack bsp stuff got merged14:01
*** gary_perkins_ has joined #baserock14:09
*** gary_perkins has quit IRC14:10
straycatwhat's the url for the ci?14:11
petefothstraycat: https://mason-x86-64.baserock.org/ ?14:22
ssam2I think the login problems on wiki.baserock.org are https:// related14:35
ssam2perhaps we need to send them our certificate: https://www.branchable.com/tips/using_https/14:36
straycatpetefoth, thanks!14:36
straycatoh it's red14:36
petefothYes but the last error was 5 hours ago, which could mean it has been building without error for 5 hours14:38
straycatoh right14:40
nowsterssam2: "HTTPS Everywhere" strikes again14:43
straycatpetefoth, seems to be failing to build stage2-gcc14:45
* petefoth withdraws, having reached the limit of his knowledge of Baserock CI14:46
straycatpetefoth, unless it's silently building something else14:46
straycati don14:46
straycat't know,14:46
pedroalvarezstraycat: CI is currently building sed14:47
pedroalvarezwell, now cython14:47
petefothstraycat: someone hels may have a better ideaopf how to a: view and b: interpret the log from the current build.14:47
straycatoh, i couldn't see that14:47
pedroalvareznot possible to see that without access to the VM14:48
straycatahhh!14:48
* petefoth thinks he should write somethign about how to check on the progress of Baserock CI (but has not idea what to write)14:49
paulsherwoodpetefoth: code? :-)14:50
straycatit would be nice if it showed the current build in progress14:50
petefothpaulsherwood: :) Don't be silly!14:50
ssam2it'd be much better if it showed what it was currently doing. I don't think anyone is working on Mason right now though14:51
ssam2there is a V2 version out for review, which I hope to deploy at some version to try out14:52
ssam2but that version doesn't have any kind of 'now building' status page either, it just sets Verified+1 or Verified-1 in Gerrit once it completes14:52
* paulsherwood wonders if he's the only one that finds it harder to contribute to reviews now gerrit is in place14:53
ssam2I think it'd be quite annoying to not know what the Mason is currently doing14:53
ssam2paulsherwood: how do you find it harder?14:53
paulsherwoodssam2: i go to any gerrit web page, glaze over, and give up14:53
* straycat actually finds it much easier14:53
paulsherwoodssam2: on mails, i knew where i was14:53
straycati've been doing reviews inbetween tasks14:54
ssam2paulsherwood: is that because the commits are at the bottom, and all in small text?14:54
ssam2*comments, not commits14:54
paulsherwoodstraycat: yes, but you had to go through an epiphany to get there... that hasn't happened for me yet :)14:54
straycatpaulsherwood, true :)14:54
*** mdunford has quit IRC14:54
paulsherwoodssam2: it's because gerrit's web ui ****ing s**ks big cows **** and i don't care who knows that i think that :)14:55
paulsherwoodoops :-)14:55
ssam2i think ':)' is becoming overused14:55
paulsherwoodonly by me14:55
straycatthough, to be fair, i am basically going to stop making series and just stick to individual changes, because series > 3 are a bit annoying to review with it imo14:55
ssam2I think high level discussion should still happen on the mailing list14:56
bashrc_the gerrit UI makes pandas sad14:56
straycatbashrc_, it has vim keys though!14:56
ssam2code reviews in gerrit, but discussions of new features, big changes etc. on the mailing list first (or, with a link to the patch sets)14:56
* bashrc_ doesn't use vim14:56
straycatthey even added visual selection14:56
straycatbashrc_, ahh i guess that's where you're going wrong :p14:57
jmacsThere are several non-vi users here14:57
ssam2straycat: they added visual selection, but they broke regular copy and paste in the process!14:57
straycatssam2, ahh well, can't have everything >.>14:58
straycat:p14:58
straycatit's definitely much nicer to use with some of keyboard shortcuts14:59
straycatyou can hit '?' to get those14:59
*** mdunford has joined #baserock15:10
bashrc_is git-review going to be in build-essential?15:41
ssam2it's in one of the strata already,. but I forget which15:41
ssam2build-essential is for the toolchain, so it wouldn't be that one15:42
persiaIf it isn't in devel, it ought be.15:46
persiaBut I don't know which is the right strata to choose for that.15:46
ssam2it's in devel already15:51
straycatshould be in devtools15:51
ssam2in devtools, indeed15:51
persiaThere we go then :)15:52
ssam2seems that Qt 5's configure script doesn't respond to SIGKILL16:02
ssam2this has turned up several pretty major bugs in distbuild16:03
ssam2or rather, it responds to SIGKILL, but the subprocesses that it started keep running regardless16:04
straycati was about to say...16:04
straycatif that's cancelling, i would guess there's an assumption that after kill's been issues it's not still building :(16:07
straycat*issued16:07
* richard_maw harrumphs, as the only way he knows of to reliably kill everything when you want to cancel a build involves cgroups, either directly (which may possibly be removed), or via systemd, which introduces an extra runtime dependency16:10
richard_mawwhich would make it difficult to run in a chroot16:10
richard_mawyou'd have to go full container16:11
ssam2it seems I could create a process group16:11
ssam2I've not tried that yet though16:11
ssam2and yes, there's an assumption that a build stops as soon as .kill() is called16:11
straycatsorry, seemed like a reasonable assumption when i wrote that :/16:12
richard_mawconversely waiting for it to exit is also problematic16:12
richard_mawas the only way I know of for getting an asynchronous notification when a process exits is with a pipe trick, and that requires the subprocess to behave16:14
richard_mawonly in last week's lwn.net was there a discussion on a patch series that proposed to fix this by giving you a FD when you clone to start the new process, which you could select on16:15
ssam2at least I can reproduce the bug with 4 lines of shell. Anything can be fixed if you have a testcase.16:17
ssam2http://ptspts.blogspot.co.uk/2012/11/how-to-start-and-kill-unix-process-tree.html seems promising for killing the whole tree16:18
ssam2(the second example)16:18
franredssam2, I did something like that in bash, the only problem I found was if the process call other processes and they do not belongs to the group you won't kill these ones16:22
ssam2oh, do forked subprocesses not get implicitly included in the group? in that case, i guess this won't work16:24
* richard_maw wonders how setsid() factors into process hierarchy stuff16:25
ssam2the article does say 'All descendant processes of the child will be added to this process group.'16:25
franredssam2, I imagine children will belong to the group but exec processes wont?16:31
ssam2any idea how I might test that? if I 'exec sh -c "sleep 10; echo HELLO"' in a shell program, you expect the shell won't be killed if I kill the whole process group?16:32
franredssam2, check the GPID of the sleep or if it is in the process tree?16:33
ssam2i'll try that16:35
richard_mawssam2: AIUI the traditional unix API way to do this would be to have the builder call setpgrp before execve, which creates a new process group with the same ID as the pid of the new process, and then use kill(-pid) to kill that process group16:56
*** Guest58619 has quit IRC17:04
*** bashrc_ has quit IRC17:05
*** jonathanmaw has quit IRC17:05
*** edcragg has quit IRC17:05
*** zoli__ has quit IRC17:06
*** zoli__ has joined #baserock17:09
*** Zara has quit IRC18:07
*** perryl has quit IRC18:07
*** pdar has quit IRC18:07
*** jmacs has quit IRC18:07
*** kejiahu has quit IRC18:07
*** wdutch has quit IRC18:07
*** benbrown_ has quit IRC18:07
*** paulsherwood has quit IRC18:07
*** SotK has quit IRC18:07
*** bwh has quit IRC18:07
*** petefotheringham has quit IRC18:07
*** DavePage has quit IRC18:07
*** bwh has joined #baserock18:09
*** petefotheringham has joined #baserock18:09
*** rdale has quit IRC18:11
straycatssam2, does ^C not cancel anymore in distbuild?18:16
ssam2should do18:17
*** zoli___ has joined #baserock18:18
ssam2keep in mind tyrell may be running random code, as it's been being used for development by SotK and perryl18:18
straycatahh good point18:19
*** zoli__ has quit IRC18:21
*** gary_perkins_ has quit IRC18:30
*** pdar has joined #baserock18:33
*** franred has quit IRC18:37
*** zoli___ has quit IRC19:22
*** Albert has quit IRC19:46
*** ssam2 has quit IRC20:09
*** paulsherwood has joined #baserock20:57
jjardonHow do you feel about using systemd firstboot to setup root password/locales/timezone... ?21:03
jjardonI just discovered it recently21:04
*** ratmice__ has quit IRC21:05
persiaAs long as there are no changes that make it mandatory, I'm fine with it.  I still hold out hope of supporting non-systemd linux (and non-linux for that matter).21:16
*** ratmice__ has joined #baserock22:04

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!