IRC logs for #baserock for Tuesday, 2017-01-17

*** idrisa_007 has joined #baserock06:13
*** jude_ has joined #baserock08:06
*** ctbruce has joined #baserock08:25
*** toscalix has joined #baserock08:31
*** rdale has joined #baserock08:33
* paulsherwood is confused as to why only runs the validate stage, not builds09:23
paulsherwoodjjardon: any idea?09:24
jjardonpaulsherwood: remove the 'only:' part from the ci job; rigth now only master and staging branches will build09:27
pedroalvarezah! the other solution is to prefix your branchname with staging?09:28
jjardonYep, but I d only do that when you finish testing and you want to send your branch for review09:29
paulsherwoodjjardon: tvm09:30
jjardonWe configured this in this way to not abuse the CI system; suggestions to improve welcomed ofc09:30
*** tiagogomes has quit IRC09:46
*** tiagogomes has joined #baserock09:49
*** ssam2 has joined #baserock10:15
*** ChanServ sets mode: +v ssam210:15
*** locallycompact has joined #baserock10:16
*** leeming has quit IRC10:40
*** leeming has joined #baserock10:43
*** jude_ has quit IRC11:07
*** jude_ has joined #baserock11:07
benbrown_ooi, does ybd still support building from chunk morphs in source repos?11:08
* paulsherwood will need to think11:09
*** jjardon has quit IRC11:31
*** jjardon_matrix is now known as jjardon11:32
pedroalvarezhm.. I think that move was done before ybd started11:33
* paulsherwood thinks so too11:33
paulsherwoodbenbrown_: why would you want that?11:34
* paulsherwood concludes the answer is no11:34
pedroalvarezyep, migration was done in Aug 2014, ybd first commit in Oct 201411:36
benbrown_paulsherwood: for the possibility of setting up CI that can auto-update refs in definitions for source we control without worrying about having to update build instructions11:36
paulsherwoodthis would just push the problem downstream... we can't assume that downstreams will maintain morph files11:41
paulsherwoodjjardon: since our definitions have always contained overlaps, stratum assembly order has always been significant11:42
paulsherwoodmorph always followed the order in definitions. ybd used to randomise, which is how we first discovered the problem11:43
paulsherwoodremoving all overlaps would be the only way to guarantee that order doesn't matter11:44
paulsherwoodtherefore for the current (and all previous) version of definitions, order is significant even if the spec doesn't currently state that11:45
jjardonok, should we document that build-essential.morph should be always the first stratum in the system .morph file then?11:45
locallycompactor toposort consistently11:46
paulsherwoodhence i propose to close and
paulsherwoodjjardon: it's not just build-essential - overlaps can occur all the way up11:46
jjardonId close those when we document this restriction somewhere11:48
paulsherwoodthis is, and always has been, a definitions issue, not ybd :)11:49
paulsherwoodbut ok11:49
jjardonpaulsherwood: sure, but how you can explain that it used to work before we applied the /usr move patches?11:50
paulsherwoodjjardon: do you really want me to spend time researching that?11:52
paulsherwoodthe answer to 'The order of the strata in the system morphology file seems to matter' is *yes, it does matter*11:52
jjardonno, document the restriction is enough for me11:53
paulsherwoodjjardon: where, precisely, please? are we using wiki.b.o for this, or spec.b.o (and if that, what generates it?)11:58
jjardonI think spec is better:
paulsherwoodok so once a mr for spec is approved, does it automagically populate d.b.o?11:59
locallycompactIt will also deploy on staging/* so be careful12:00
pedroalvarezd.b.o is then no12:01
jjardonoh yeah, it will populate spec.baserock.org12:02
locallycompactisn't docs just pointing at spec?12:02
pedroalvarezspec.b.o was created to demonstrate that it was possible. Should we kill current docs.b.o and make it point to the same place?12:02
pedroalvarezor do we prefer spec?12:03
locallycompactmerge docs and spec and point to the same I think12:03
jjardonlets keep but pointing to spec.baserock.org12:04
paulsherwoodwhat about also12:10
locallycompactwiki.baserock is crazy out of date12:10
pedroalvarezI think it has different use12:11
paulsherwoodagreed. it's probably time for a spring clean :)12:11
paulsherwoodpedroalvarez: it does, but it's out-of-date :)12:11
*** paulwaters_ has joined #baserock12:11
locallycompactstill talks about setting up a build environment via baserock vm12:12
pedroalvarezI'm not sure it's worth starting documenting things now :)12:12
paulsherwoodjjardon: wrt - it seems to only happen on gitlab... any idea what's special?12:13
jjardonno idea; the runners are a pristine docker image (python:2.7-slim to be exactly) in every build12:14
pedroalvarez DNS record updated, might take some time to propagate12:40
pedroalvarezI just noticed that now it's blue12:40
paulsherwoodwow :)12:53
jjardon"Your connection is not secure" if I try to go to
pedroalvarezplain http14:11
pedroalvarezI don't know yet how I feel about uploading private SSL certs to Gitlab14:12
pedroalvarez(to make https work)14:12
ssam2something probably needs changing in the haproxy config, if is still going through haproxy14:14
ssam2http:// redirects to https:// at the moment for docs.b.o14:14
pedroalvarezit's not going through the haproxy anymore. To point to the gitlab page docs. is now a CNAME pointing to ""14:21
pedroalvarezhm.. using the haproxy might be a better idea, yes14:21
*** gtristan has joined #baserock15:09
*** ctbruce has quit IRC15:39
*** ctbruce has joined #baserock15:43
*** ctbruce has quit IRC16:24
*** Guest77985 has quit IRC17:33
*** jude_ has quit IRC18:05
*** locallycompact has quit IRC18:13
*** tiagogomes has quit IRC18:39
*** ssam2 has quit IRC18:45
*** jude_ has joined #baserock19:04
*** toscalix has quit IRC20:20
*** rdale has quit IRC20:37
*** gtristan has quit IRC21:26
*** jude_ has quit IRC21:47
*** jude_ has joined #baserock21:59
*** jude_ has quit IRC22:13
*** jude_ has joined #baserock23:31

Generated by 2.15.3 by Marius Gedminas - find it at!