IRC logs for #baserock for Thursday, 2016-06-09

*** edcragg has quit IRC00:14
*** gtristan has quit IRC03:03
*** gtristan has joined #baserock03:59
*** lachlanmackenzie has quit IRC05:23
*** lachlanmackenzie has joined #baserock05:24
*** jjardon has quit IRC05:25
*** jjardon has joined #baserock05:34
*** CTtpollard has quit IRC07:24
*** CTtpollard has joined #baserock07:28
*** toscalix has joined #baserock07:53
*** edcragg has joined #baserock08:04
*** edcragg has quit IRC08:14
*** tiagogomes has joined #baserock08:21
*** tiagogomes has quit IRC08:33
*** rdale has joined #baserock08:34
*** tiagogomes has joined #baserock08:37
*** gtristan has quit IRC08:40
*** edcragg has joined #baserock08:41
*** jonathanmaw has joined #baserock08:46
*** tiagogomes has quit IRC08:48
*** tiagogomes has joined #baserock08:50
*** tiagogomes has quit IRC09:07
*** tiagogomes has joined #baserock09:09
*** locallycompact has joined #baserock09:43
*** locallycompact has quit IRC09:44
*** locallycompact has joined #baserock09:45
*** gtristan has joined #baserock09:51
paulsherwoodfolks, can anyone shed any light on what might be happening in
paulsherwoodi'm currently thinking this is to do with the system ybd is running on, rather than ybd or sandboxlib10:47
CTtpollardis it to do with python2.7-3?10:48
SotKpaulsherwood: what python version are you using?10:50
SotKah, 2.710:51
* SotK is blind10:51
locallycompactpaulsherwood, are you trying to build a 64 bit system on a 32 bit runner10:52
paulsherwoodlocallycompact: yes, i am10:52
paulsherwoodwhy should that matter?10:52
locallycompactIsn't that impossible?10:52
paulsherwoodoh... would the reverse also be impossible?10:53
paulsherwoodie could i build 32-bit on a 64-bit runner?10:53
locallycompactthe reverse should not be impossible, but it happened to not work when we tried it originally10:53
paulsherwoodah, ok...10:53
jmacsWhy is 64-on-32 impossible, out of interest?10:54
rjekThat's cross compiling :)10:54
SotKisn't this error the one described here?
jmacsoh yes, we don't do that, do we10:55
persiaTechnically, 32-bit on 64-bit is also cross-compiling, but for some 64-bit implementations (aarch64, ppc64, x86_64), there are tricks one can do to make the runner pretend to be 32-bit so it isn't cross-compiling.10:57
rjekAlso works on SPARC10:59
rjekOr UltraSPARC I should say10:59
persiarjek: Does it?  I had trouble with it some time in the past.11:01
* persia is glad to hear of progress.11:01
rjekEverybody who has ever used SPARC systems has had trouble with them.11:01
persiaI also heard it didn't work for MIPS, but I haven't tried that one.11:01
persiaThe only trouble I had with SPARC was linux32: everything else seemed to work fine.  Maybe it was just me.11:02
rjekMIPS64, IIRC, is exactly the same as MIPS32 except the registers are twice as wide; 32 bit code should "just work" in 64 bit mode due to some tricks11:02
persiaYes, but is there `linux32` support?11:02
rjekI think so; although I seem to recall it's implemented as an LD_PRELOAD on some systems.11:03
persiaLooking at the source, it *should* work for MIPS and AXP, but I don't have any AXP anymore, and my usual MIPS target isn't MIPS64, so I can't verify.11:05
persiaAlso s390/s390x11:05
paulsherwoodSotK: i guess you are right11:11
locallycompactI pushed the data model we were talking about last week in disambiguating "chunk files" into what they actually are, which is "build templates"11:16
locallycompactparses this
paulsherwoodoh, that makes sense!11:30
* SotK thinks its much better than the current state of things11:32
paulsherwoodi'd go a bit further, though... s/chunk/contents/ and have that rust can itself have repo, ref, bit11:32
paulsherwoodthis would allow us to deal with stratum: ansible contains chunk ansible and a load of other things required to build ansible11:32
SotKI'm confused as to the benefit there11:33
SotKyou can do that with the example given surely?11:33
paulsherwoodthe example has rust-lang and rust, which avoids the problem iiuc11:34
paulsherwoodi want unique names... hence ansible contains ansible is not good for me11:34
paulsherwoodand zookeeper contains zookeeper etc11:35
locallycompactrust-lang was just to reflect the name of the repository. if I called it "rust" I wouldn't have a problem in haskell because Chunk and Stratum are different types11:36
paulsherwoodwe have more than 10 examples of this11:36
paulsherwoodso you don't want to go for generic nesting? you still want explicitly chunk, stratum, system, cluster?11:36
* paulsherwood reminds locallycompact that names are important for people, whether the code cares about them or not11:39
SotKpaulsherwood: aha, that makes sense11:40
locallycompactWe always say "the ansible stratum", "the ansible chunk"11:40
locallycompactbut that's *because* they're loosely typed like that11:40
SotKlocallycompact: how would you know which to build if I did `build-tool rust x86_64`?11:40
paulsherwoodSotK: exactly :)11:41
locallycompactI'd say that doing that for a chunk is already ill-defined currently and this hasn't made a difference11:41
locallycompactTwo chunks in different stratum with the same name is already ambiguous, the only way to distinguish it is to force qualification, stratum.chunkname11:41
paulsherwoodit has made a difference. every little cranny of uncertainty needs to be cleaned out11:41
paulsherwoodthis would be solved tidily by what i said above, ie11:42
paulsherwood12:32 < paulsherwood> i'd go a bit further, though... s/chunk/contents/ and have that rust can itself have repo, ref, bit11:42
locallycompacts/chunk/contents doesn't do much for the types11:43
paulsherwoodi want generic nesting... but others may disagree11:44
paulsherwoodin my dream scenario, contents is a list of definitions... each definition may have further contents11:45
locallycompactCan you sketch out an example yaml11:45
locallycompactI don't think it's unreasonable to want any concievable grammar representation tbh.11:46
locallycompactIn fact I welcome that problem11:46
paulsherwoodi'll have a think11:51
*** tiagogomes has quit IRC12:32
*** tiagogomes has joined #baserock12:58
leemingIf I've written some lorry files. How do i test them before submitting a merge request?13:09
leemingor is there a build system similar to in gitlab?13:10
SotKyou can run lorry locally passing it the lorry file you made13:10
* leeming scratches head13:15
leemingok but doesnt that mean running a local trove?13:15
SotKnope, but it is a little annoying if you aren't in a baserock devel system13:16
* SotK finds links13:16
SotKyou can clone lorry from and run it from the checkout13:17
SotKbut it depends on various version control systems iirc13:18
leemingoh right.. i have that checkedout13:19
SotKok, I think you can do `./test-lorry /path/to/lorry`13:20
SotKha, or not13:21
SotK`SRCDIR="." ./test-lorry /path/to/lorry` seems better13:22
* leeming actually didnt have lorry.git checkouted out, actually had lorrieS.git , was confused for a sec13:22
leemingSotK, correct to assume if the script returns empty, the file is correct?13:39
SotKI expect so, do you have a 'workd' directory now? (and if so, does it contain the thing you tried to lorry?)13:40
leemingah yes, thanks13:41
leemingok I have some additional lorry files. What is the process for submitting a patch/merge request13:48
SotKdo you have a gerrit account yet?13:49
leemingeerrr, this is baserock, so i guess not...13:50
leemingwill make one now13:50
* leeming wonders if email is being that slow.. or if registration failed13:57
leemingpedroalvarez, i didn't mess up my email AGAIN did i? :\14:18
leemingstarting to feel like one of them days14:18
SotKwhen you've got your account set up, you'll probably want to add your ssh key to it on the preferences page in Gerrit14:34
SotKthen do `git review -s` in the checkout of the repo you are sending patches to14:36
SotKand `git review` to actually send the patch14:36
*** toscalix has quit IRC15:10
*** toscalix has joined #baserock15:11
*** toscalix_ has joined #baserock15:15
*** toscalix has quit IRC15:15
*** jonathanmaw has quit IRC16:31
*** toscalix_ is now known as toscalix16:32
*** rdale has quit IRC16:59
*** edcragg has quit IRC17:10
*** edcragg has joined #baserock17:37
*** toscalix has quit IRC17:42
*** edcragg has quit IRC17:56
*** locallycompact has quit IRC18:01
* leeming has been out of the office for a while on an errand18:02
leemingconfused me for a sec there. git-review is not vanilla git ;)18:03
leemingERROR: missing Change-Id in commit message footer18:07
leemingshould i blindly follow the example given? i.e. gitdir=$(git rev-parse --git-dir); scp -p -P 29418 ${gitdir}/hooks/18:07
*** jjardon has quit IRC18:12
*** jjardon has joined #baserock18:17
*** jjardon has quit IRC18:22
*** jjardon has joined #baserock18:23
persialeeming: not blindly, but basically, yes.18:27
leemingim just reading right now what this all is18:28
leemingreview process is different to what im using on my main project :S18:28
leeming"There's a git add-on called git-review that adds a Change-ID line to your commits and manages other aspects of using Gerrit."18:37
leemingok well patch submitted now. jobs a good'un, hopefully i didn't screw up any more than I already have today.19:33
leeminghome time19:33
*** cosm has joined #baserock23:24

Generated by 2.15.3 by Marius Gedminas - find it at!