IRC logs for #baserock for Wednesday, 2016-02-10

*** gtristan has quit IRC07:14
*** gtristan has joined #baserock07:42
*** toscalix has joined #baserock07:59
*** cosm has quit IRC08:38
*** bashrc has joined #baserock09:01
*** ctbruce has joined #baserock09:08
*** edcragg has joined #baserock09:10
*** edcragg has quit IRC09:27
*** rdale has joined #baserock09:33
*** jonathanmaw has joined #baserock09:41
*** edcragg has joined #baserock09:57
*** gtristan has quit IRC10:01
*** ssam2 has joined #baserock10:10
*** ChanServ sets mode: +v ssam210:10
*** Lachlan1975 has joined #baserock10:25
benbrown_richard_maw: I'm actually a little confused by your reply to my ybd patch re remote.origin.mirror, `man git config` doesn't mention that behaviour?10:29
benbrown_I may have misunderstood what you meant though.10:30
benbrown_As far as I'm aware, --prune would need to be added to both fetch and remote update to delete refs that no longer exist?10:30
richard_mawmirror mode is a bit magic, I'll see if I can find some documentation to point you at in a bit10:31
benbrown_richard_maw: hehe, thanks :)10:32
*** Lachlan1975 has quit IRC10:40
*** locallycompact has joined #baserock10:50
*** gtristan has joined #baserock10:54
richard_mawbenbrown_: turns out remote.origin.mirror is actually just about push behaviour, which will delete on the remote end10:55
richard_mawbenbrown_: so it was entirely redundant in the original config since we never push from our git cache10:56
* richard_maw had the behaviour on push confused with the behaviour on fetch, since it would have been logical for it to be symmetric10:56
benbrown_richard_maw: Cool, thanks for clearing that up :)10:59
*** ctbruce has quit IRC12:03
*** ctbruce has joined #baserock12:17
pedroalvarezlocallycompact: just curious, do you know that we have gerrit available for code review?12:20
locallycompactpedroalvarez, yes, and sorry but once was enough12:20
pedroalvarezhaha fair12:21
pedroalvarezpatches in the mailing list are welcome too12:21
* paulsherwood remembers the days when the mailing list was full of patches12:24
paulsherwoodusing gerrit has a side-effect of making the ml seem much quieter than it was12:24
paulsherwoodand we lose some feedback that people would give when seeing ml patches12:25
* paulsherwood does not see any gerrit patches by default, for example, therefore does not comment on them12:26
pedroalvarezlocallycompact: have you actually used this ansible script? there is something that I'm not sure is syntactically correct12:27
*** ctbruce has quit IRC12:27
locallycompactI have, what is wrong?12:27
pedroalvarezthe "get Riemann dash gems" task12:28
pedroalvareznote, I said "I'm not sure"12:28
*** rdale_ct has joined #baserock12:29
locallycompactOh, is it that you can't do multiple gems?12:29
*** rdale has quit IRC12:30
pedroalvarezwell, you can but I've never seen that syntax anywhere.12:39
pedroalvarezI've answered12:39
locallycompactah,  I see12:43
locallycompactthe gems weren't actually working12:43
locallycompactas it was12:44
*** gtristan has quit IRC14:03
*** bashrc has quit IRC14:15
*** bashrc has joined #baserock14:21
paulsherwoodJPohlmann: is there a visible demo instance of custard anywhere? (yes, i'm stalking you...)14:33
*** gtristan has joined #baserock14:40
*** bashrc has quit IRC14:57
*** bashrc has joined #baserock14:57
locallycompactHas anyone used AWS to spin up a baserock vm? The launch instance options don't seem to allow for a custom image - just redhat and ubuntu things15:13
paulsherwoodiirc it was previously possible to use custom images... but i failed when i tried15:13
richard_mawlocallycompact: technically I have, couldn't get it to work though15:13
paulsherwoodrichard_maw may know more15:14
* richard_maw will try to find where he wrote up his steps15:14
paulsherwoodrichard_maw: oh, did you succeed, then?15:14
richard_mawpaulsherwood: probably not, but without any way to get output from the machine it was impossible to tell15:15
paulsherwoodso you couldn't ssh in, i assume15:15
richard_mawlocallycompact: I'll PM you the link to where I wrote it up, since it's not recorded in a place for public consumption15:16
richard_mawpaulsherwood: that, and no serial available15:16
richard_mawthere was some evidence the machine claimed the IP address it was allocated over DHCP15:16
richard_mawlocallycompact: the gist of the steps to get as far as I did was: 0. make a disk image, 1. import the volume into AWS, 3. convert it into a snapshot 4. register the snapshot as a virtual machine15:18
*** CTtpollard has quit IRC15:57
*** jonathanmaw_ has joined #baserock15:58
*** jonathanmaw has quit IRC16:01
*** jonathanmaw_ is now known as jonathanmaw16:05
*** CTtpollard has joined #baserock16:07
*** faybrocklebank has quit IRC16:10
*** ctbruce has joined #baserock16:34
*** tiagogomes has quit IRC16:37
*** edcragg has quit IRC16:39
*** CTtpollard has quit IRC16:41
*** edcragg has joined #baserock16:41
*** CTtpollard has joined #baserock16:42
*** jonathanmaw_ has joined #baserock16:44
*** jonathanmaw has quit IRC16:45
locallycompactIs the go that's already on baserock lorrying properly? It only goes up to 1.416:46
pedroalvarezdelta/go.git ?16:46
* pedroalvarez goes to have a look16:47
*** ctbruce has quit IRC16:49
*** tiagogomes has joined #baserock16:49
pedroalvarezlocallycompact: there is something weird about it... it's an hg lorry from
pedroalvarezbut that now links to a github repo o.016:50
* richard_maw claps16:50
* pedroalvarez too16:50
richard_mawhand to forehead clapping, rather than hand to hand16:50
JPohlmannpaulsherwood: There isn't, nope16:50
paulsherwoodJPohlmann: ah, ok16:51
* paulsherwood can wait :)16:51
JPohlmannpaulsherwood: I could spin up a demo somewhere16:51
pedroalvarezaaaand... latest commit of go in gbo is not present in the github repo16:51
pedroalvarezmoar clapping16:52
paulsherwoodpedroalvarez: not surprising, their conversion will be different from ours16:52
paulsherwoodwe'll need a new go-git, i expect16:52
JPohlmannpaulsherwood: ...if you remind me over the weekend? I'm pretty busy getting started with customers during the week16:52
paulsherwoodJPohlmann: don't worry - i don't want to create work for you. how 'mature' is it?16:53
jjardonHi! Can I have another review for ? python support is broken for some modules without that patch16:54
JPohlmannpaulsherwood: Not very. The UI has a few issues and there is currently no useful data validation. And I've only recently started to look into a decent history feature.16:55
JPohlmannAt this point Mustard is more feature-complete.16:55
richard_mawjjardon: I strongly believe that all python3 things shouldn't symlink themselves into where python2 things put themselves16:56
richard_maw$PREFIX/bin/python{,-config} should be for python2 IMO16:57
paulsherwoodJPohlmann: ack16:57
jjardonrichard_maw: Do you think the bug is in the chunks then? They are expecting python-config to exist16:57
richard_mawjjardon: I think if they are asking for python-config, they want python216:58
richard_mawand if they want python, they want python216:58
jjardonrichard_maw: no, its a python3-compatible chunk; It can be a bug use python-config unconditionally though16:59
richard_mawif it's asking for python-config it should get python2-config, if not, yes, it's a bug in its python3 compatibility16:59
jjardonrichard_maw: are you suggesting that python should not exist in a python3-only system? I have no problems with that but I think thats agains one of the python PEPs17:00
rjekSounds like a packaging/distro issue, not a python issue17:00
* richard_maw would have to re-read the pep17:01
richard_mawbut the last time I read one like that, I came to the conclusion that you should not have python if it is not python217:01
* jjardon rechecks17:01
richard_mawthere was some mumbling about "python" being only for the subset of python that is interpretable identically in both17:02
richard_mawbut given the amount of existing software that assumes "python" == "python2", to try that would be antisocial17:02
richard_mawindividual applications may decide to play nice with specifying python for the intersection of the two17:03
richard_mawbut as a distro packager, unless you have the ability to enforce that only software which conforms to that convention is included17:03
richard_mawyou have to say python == python217:04
* richard_maw steps down from his soapbox17:05
jjardonrichard_maw: ok, I will actually change the patch to not generate a "python" symlink and see what breaks17:06
* richard_maw recalls that causing python2 to misbehave, since it assumed that it shouldn't stomp over the /usr/bin/python symlink with python2 if it existed at build time17:07
richard_mawwhich was another reason why I didn't like the python -> python3 symlink17:08
locallycompactwhat does this indicate? (ybd)
paulsherwoodalmost certainly, that i'm an idiot. which version of ybd?17:10
locallycompactI think it was 16.01 with some hotfixes, let me try master17:10
paulsherwoodlocallycompact: to avoid me constantly asking that question, you could paste the whole log :)17:10
jjardonrichard_maw: python symlink will be rewriten if python2 is installed:
richard_mawjjardon: aye, but if you hadn't made python3 install the python symlink, you wouldn't have needed to add that post-hoc, after much swearing at it breaking things when both python2 and python3 were installed17:11
paulsherwoodlocallycompact: hmm. could you try with log-verbose: True, please?17:14
jjardonrichard_maw: that line was changed at the same time the python3 transition was made ... if something broke was because some missing python2 dependencies maybe?17:17
paulsherwoodlocallycompact: looks like it's not actually doing anything in the go1 sandbox, which is weird. does your go1 definition have a repo: and ref: ?17:19
paulsherwoodyes, it does.17:20
paulsherwoodwhat about a build-system?17:20
locallycompactno, should it?17:21
jjardonrichard_maw: :) our own python-distutils instructions will break if "python" doesn't exist17:21
*** jonathanmaw_ is now known as jonathanmaw17:22
richard_mawjjardon: yeah, but if you want python3 of those, you should really be explicit17:23
paulsherwoodlocallycompact: this is still a bug in ybd (should make it more obvious what's happened), but yes, better to specify what build-system you expect it to use, unless you've got a morph file specifying the build-steps explicitly17:23
richard_mawjjardon: since there's a little magic in, in that it installs the packages for the version of python that invoked the command17:23
richard_mawjjardon: I'd rather you explicitly had it run `python3 install`, and have a separate chunk with a different name, to prevent confusion when people try to use python modules for the wrong version of python17:24
*** jonathanmaw has quit IRC17:24
locallycompactpaulsherwood, I do have a go/go1.4.3.morph with build steps17:24
locallycompactshould that be enough?17:24
*** jonathanmaw has joined #baserock17:24
jjardonrichard_maw: should we have python3-distutils in DEFAULTS, then ?17:24
richard_mawjjardon: in the absence of parameterised builds, yes17:25
jjardonrichard_maw: rigth17:25
paulsherwoodlocallycompact: yes, it should. i'm confused, then17:25
locallycompactsetting build-system: manual in the chunk part of strata morphology seemed to get further17:26
pedroalvarezlocallycompact: build-system shouldn't be there if you are also creating  a .morph file with the instructions17:27
pedroalvarezit is redundant, and can create conflicts17:28
richard_maw^ yaml document17:28
locallycompact^ mammal document17:29
paulsherwoodlocallycompact: in any case, would you mind publishing your definitions? i've not seen this issue before17:29
*** jonathanmaw has quit IRC17:29
* paulsherwood would like to reproduce, and close the hole17:30
rdale_ct'manual' is a set of empty build instructions in the defaults17:30
locallycompactsure let me just do it all once more17:30
*** tiagogomes has quit IRC17:34
jjardonmmm, its a bit weird I have to patch morph to support a new build system in DEFAULTS; paulsherwood, does ybd have the same 'problem' ?17:46
paulsherwoodi believe ybd should read DEFAULTS from definitions, will fall back to its own copy if none present17:47
pedroalvarezjjardon: uh? isn't morph using them from DEFAULTS?17:49
pedroalvarezI mean, why is patching morph needed?17:49
jjardonpedroalvarez: I think because this line
jjardonthe build system should exist in (I can be wrong)17:52
pedroalvarezI can't find anything obvious in the line you posted17:55
*** ssam2 has quit IRC17:55
paulsherwoodlocallycompact: i'll have to investigate this overnight, sorrty17:58
*** bashrc has quit IRC18:03
*** locallycompact has quit IRC18:05
rdale_cti think the go build problem might be because ybd doesn't set up TMPDIR and it is defaulting to /var/tmp which might not be writable18:12
jjardonI think ybd should use $XDG_CACHE_HOME, not TMPDIR (wich is notmally /tmp, hence a tmpfs filesystem)18:17
jjardonrichard_maw: patches to add the new python3-distutils build system sent:
*** edcragg has quit IRC18:18
pedroalvarezjjardon: nice to see that patching morph wasn't needed18:20
jjardonpedroalvarez: patch coming for morph as well ;)18:21
jjardon(ybd works without patching btw)18:21
pedroalvarezmorph should as well18:21
jjardonpedroalvarez: do the test: try to build the branch Ive just submitted18:22
pedroalvarezthat patch shouldn't be merged until we do a release with Morph fix in then18:22
pedroalvarezyeah, I trust you, but it shouldn't fail IMO18:22
jjardonyep, I think its really a bug in morph18:23
pedroalvarezKeyError: 'Unknown build system: python3-distutils'18:27
*** toscalix has quit IRC18:41
pedroalvarezI think I found the error, but it's going to be tricky to fix18:53
*** locallycompact has joined #baserock19:02
pedroalvarezcreated a story for it:!/story/7219:03
paulsherwoodrdale_ct: ybd does setup TMPDIR now -
paulsherwoodi tried honouring $XDG_CACHE_HOME instead of $base/tmp by default, but it broke for some users19:05
rdale_ctif that TMPDIR change is recent, maybe it isn't in the version of ybd that locallycompact is using19:27
locallycompactI have that change19:30
rdale_ctah ok19:30
rdale_ctnot that then19:30
rdale_ctthis won't do anything as a configure command: export GOPATH=$(pwd) as the env variable value will be lost when running the build command19:33
locallycompactyeah I expected as much19:34
*** edcragg has joined #baserock20:30
paulsherwoodso i've made ybd explicitly error and raise an exception when it fails to autodetect buildsystem20:38
paulsherwoodhopefully that's clearer in future20:38
paulsherwoodtechnically definitions now are supposed to insist on buildsystem being defined, iirc, so ybd is misbehaving by even attempting autodetect :)20:39
paulsherwoodlocallycompact: i assume you've seen
paulsherwoodthat's how i got pre 1.4 go to work20:41
*** cosm has joined #baserock20:55
locallycompactpaulsherwood, I hadn't20:57
paulsherwoodah, ok :) well... maybe it will help :-)20:57
locallycompactpaulsherwood, what do I put for build-system, manual?20:58
paulsherwoodstretch goal might be to add a go-lang to defaults, including indicators :)20:59
paulsherwoods/go-lang/go-lang buildsystem/20:59
paulsherwoodlocallycompact: you remember your error from the other day, when i mentioned about the warnings because missing .unpacked files? did the problem go away after that was fixed21:03
locallycompactCan't remember which one that was right now21:04
locallycompactOh yeah that fixed it21:05
*** gtristan has quit IRC21:19
paulsherwoodkbas now serving over 10,000 artifacts :)21:31
*** bfletcher has quit IRC21:42
*** zoli_ has quit IRC21:42
*** franred_ has joined #baserock21:42
*** franred has quit IRC21:48
*** trn has quit IRC21:48
*** lachlanmackenzie has quit IRC21:48
*** lachlanmackenzie has joined #baserock21:57
*** trn has joined #baserock22:21
*** zoli_ has joined #baserock22:36
*** cosm has quit IRC22:39
*** bfletcher has joined #baserock23:03
*** edcragg has quit IRC23:39

Generated by 2.15.3 by Marius Gedminas - find it at!