IRC logs for #baserock for Tuesday, 2016-08-02

*** anahuelamo has quit IRC01:47
*** persia_ has quit IRC05:28
*** persia_ has joined #baserock05:31
*** gtristan has quit IRC06:17
paulsherwoodi'm looking at https://github.com/devcurmudgeon/ybd/issues/23307:11
paulsherwoodand notice that i can fix it by using sh -c -e, rather than just sh -c. iirc i borrowed this from morph. is there any reason not to have -e here?07:13
paulsherwood(where 'here' is the invocation of *-commands in sandbox)07:16
perrylpaulsherwood: iirc the -e flag just sets the command to immediately exit on error without reporting07:17
paulsherwoodyes, and that's ok behaviour i think.07:18
perrylactually yeah that'd be fine if the intent is to stop it crashing on error i think07:21
paulsherwoodthe intent is to stop on error07:22
paulsherwoodit currently doesn't, for this specific situation (set of commands in a block)07:22
*** CTtpollard has joined #baserock07:23
*** toscalix has joined #baserock07:33
*** gtristan has joined #baserock07:58
* paulsherwood wonders why upgrading a ubuntu vm now requires user to specify a 'chef url'08:05
* gtristan is getting bitten by trying to do things properly... so here is a lorry problem; if I do svn checkout svn://svn.code.sf.net/p/openil/svn ... I get an svn/ directory with trunk, branches and tags directory... looks like a standard layout08:11
gtristanwhen I lorried it, it gave me something different :-/08:11
paulsherwoodeek08:12
gtristanthis is the upstream svn https://sourceforge.net/p/openil/svn/HEAD/tree/08:12
gtristanand this is the lorry file: https://paste.fedoraproject.org/399658/70125570/08:13
gtristanI'm currently doing, by hand, what lorry seems to intend to do, i.e. git svn init, git config (stuff for trunk, branches and tags)08:14
gtristanand currently running git fetch08:14
gtristanthat git fetch seems, so far to be building the right thing08:15
gtristan(it's creating some branches and tags which make sense / match the expected branches and tags)08:15
gtristanmaybe it's a case of a really old version of lorry running there08:16
*** vgrade12 has quit IRC08:17
* paulsherwood can't see anythign wrong with the lorry file08:18
* gtristan is again running lorry manually08:20
gtristanI did that yesterday, and it was in the middle of building the right thing (but did not yet complete), so I pushed up the lorry file thinking it would work properly08:20
locallycompactgtristan, if that's running on a downstream trove it's going to collide with the version of devil that's already there08:21
* gtristan knows it was the right thing just from cd'ing into the lorry workd directory and typing git branch -r and git tag08:21
*** anahuelamo has joined #baserock08:21
gtristanlocallycompact, umm, I dont believe there is one already there08:22
paulsherwoodhttp://git.baserock.org/cgit/?q=devil08:22
* gtristan grepped (insensitive/recursive) his checkout of baserock's lorries for 'devil' and found nothing 08:22
*** mario-goulart has left #baserock08:23
gtristaneek08:23
paulsherwoodrecent addition i think08:23
gtristanmust be fairly recent yeah08:24
paulsherwoodDaniel Firth 2016-07-21 12:26:3508:24
*** locallycompact has quit IRC08:25
paulsherwoodanyway, i guess that's usable for you?08:25
gtristanpaulsherwood, no08:29
gtristanpaulsherwood, I renamed the add to devil-svn (file and name inside lorry) so that it doesnt clash anymore08:29
paulsherwoodack08:29
paulsherwoodso there are two devils?08:29
paulsherwooddevil upstreams?08:29
gtristan(the github repo may be some fork or mirror, but when creating that github they failed to import the relevant tag/branch information)08:30
paulsherwoodah,ok08:30
gtristanthe github page doesnt really say much, but it does refer to the sourceforge project page08:30
gtristanso maybe they are continuing development on github, not sure08:30
*** CTtpollard has quit IRC08:40
*** CTtpollard has joined #baserock08:41
*** jonathanmaw has joined #baserock08:48
*** franred has joined #baserock08:52
*** locallycompact has joined #baserock09:24
*** gtristan has quit IRC12:20
*** gtristan has joined #baserock12:29
pedroalvarezso, delta/devil.git is  unusable?12:33
pedroalvarezI suggested using that github mirror because it's being referenced in their homepage? http://openil.sourceforge.net/12:34
pedroalvarezI would suggest trying to use github's one12:41
pedroalvarezRelated question: Are we lorrying things in upstream trove for downstream definitions?12:42
persiaI believe there are a couple cases where a downstream project asked for an upstream trove mirror.12:55
persiaBut I don't think it is general practice: I think most downstream troves mirror their own additional stuff, so as not to have to coordinate with upstream lorry configuration.12:56
persiaThat said, there are a number of users who have downstream definitions and use the upstream trove, so need things mirrored in upstream trove to have a mirror.12:56
pedroalvarezI've submitted a few things:14:01
pedroalvarezhttps://gerrit.baserock.org/#/c/2268/114:01
pedroalvarezhttps://gerrit.baserock.org/#/c/2269/114:01
pedroalvarezhttps://gerrit.baserock.org/#/c/2270/114:01
franredpedroalvarez, that build of wpa_supplicant is very suspicious...14:06
pedroalvareznote that I'm not modifying anything14:06
pedroalvarezthat configuration file was generated in this branch: http://git.baserock.org/cgit/delta/hostap.git/log/?h=baserock/genivi/morph14:06
franredso you have added the configuration file instead of leave the system to build one on the fly?14:08
pedroalvareznope, I'm adding the configuration file in the chunk definition instead of carrying a patch14:09
pedroalvarezdoes that make sense franred?14:18
pedroalvarezbetter to not carry paches, imo14:18
persia+1 on not carrying patches14:24
*** rdale has quit IRC14:26
franredpedroalvarez, yes, make sense14:26
pedroalvarezkewl14:27
*** rdale has joined #baserock14:32
*** anahuelamo_ has joined #baserock14:44
*** anahuelamo has quit IRC14:44
*** edcragg has quit IRC15:08
*** edcragg has joined #baserock15:16
*** faybrocklebank has quit IRC15:21
paulsherwoodanyone know what's special about nfs? i'm being told about strange behaviour running ybd on nfs build machines...16:10
edcraggstorage on nfs? aren't there some permissions quirks16:11
richard_mawplus some operations aren't guaranteed to be consistent; generally it's not POSIX compliant16:11
richard_mawits flock() implementation has classically been totally broken16:12
paulsherwoodeek, then16:12
paulsherwoodso one example problem is 'tar: ./usr/src: Cannot mkdir: No space left on device'16:13
paulsherwoodon a build machine that notionally has 70GB+ free at start16:13
richard_maw70GB+ on the nfs share?16:15
paulsherwoodseems so16:15
paulsherwoodi was wondering if somehow the chroot could get confused on nfs16:16
edcraggnot sure how free space calculation works on nfs, but it may differ from local storage16:18
paulsherwoodack16:33
*** jonathanmaw has quit IRC17:04
*** locallycompact has quit IRC17:16
*** toscalix has quit IRC17:33

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