IRC logs for #baserock for Wednesday, 2015-07-08

*** sherm_ has joined #baserock06:46
*** mariaderidder has joined #baserock06:59
*** franred has joined #baserock07:07
*** rdale has joined #baserock07:20
*** bashrc_ has joined #baserock08:05
*** jonathanmaw has joined #baserock08:26
*** ssam2 has joined #baserock08:54
*** ChanServ sets mode: +v ssam208:54
*** lachlanmackenzie has joined #baserock09:08
*** sherm_ has quit IRC09:09
*** sherm_ has joined #baserock09:24
ssam2paulsherwood: as someone who prefers github workflow over gerrit, this might be interesting if you've not seen it: http://gogs.io/10:23
KinnisonGogs, as with other attempts (github, gerrit, gitlab, etc) suffers the (IMO Fatal) flaw of not storing its data in git10:25
KinnisonAlso, Gogs is written in a fashion statement of a language.10:25
petefothKinnison: Haskell?10:26
pedroalvarezGo I believe10:27
petefoth:)10:27
bashrc_Gogs works pretty well. Currently I use it as a backup system, assuming that Github will eventually turn into Sourceforge10:29
Kinnisonpetefoth: haskell isn't popular enough to be a fashion statement10:29
*** mariaderidder has quit IRC10:30
jmacsI don't care what GitHub is written in; it works, so I've no need to fix it.10:30
jmacsIf Gogs works, the same will apply10:31
bashrc_Github serves its purpose as a relatively user friendly interface for basic project management10:31
bashrc_if you can use Github then you can use Gogs. There's little difference10:32
*** jonathanmaw has quit IRC10:36
*** cosm has quit IRC10:44
*** mariaderidder has joined #baserock10:45
*** jonathanmaw has joined #baserock11:04
paulsherwoodssam2: i saw that. i'm coming around to gerrit, tbh.11:19
* paulsherwood would prefer gitano + workflow though11:19
paulsherwoodjmacs: github works for projects up to a certain scale/complexity... not so good for big integrations imo11:20
paulsherwoodor big teams for example11:20
jmacsThat may well be true; I've seen it work for bigger teams than baserock, though11:22
paulsherwoodfair11:22
paulsherwoodbut what do you mean by baserock, precisely? i don't think gh could be used for what gbo does?11:23
persiaI think it is more about the number of simultaneous changes thatare in flight than strictly about team size.11:24
jmacsNo, I wouldn't try to lorry things in github.11:24
paulsherwoodpersia: agreed, and i would hope to see a big increase in changes, given how fast things change upstream11:25
jmacsI would perhaps keep definitions and morph on there.11:25
paulsherwoodon gbo?11:25
jmacsOn GitHub.11:25
SotKWhat about the other things that are currently on Gerrit?11:27
pedroalvarezI assumed that "definitions and morph" means all of the baserock/* projects11:28
paulsherwoodheh... just as i was considering asking for ybd to move from gh to gbo/gerrit... folks are thinking about moving in the other direction :011:29
jmacsI'm not suggesting moving an established project, just that I would consider github valid if a choice was being made.11:29
pedroalvarezI don't feel like moving from gerrit to another thing :)11:29
* SotK neither11:30
persiaI'd be +1 for moving from gerrit if something else had similar CLI tooling and had the git-reliable feature so that we didn't need to maintain two git repos.  Not for any other reason, probably.11:30
* persia finds the sync issues between gerrit and gitano annoying11:31
SotKpersia: Which sync issues are they? I thought gerrit -> gitano was almost instant?11:33
persiaSotK: It's the "almost".11:33
persiaWe've had a number of issues when folk merged things and pushed to gitano.11:33
persiaAnd we have a bunch of automation to articulate the relationship, which is somewhat fragile.11:33
ssam2we could turn it into one-way sync11:42
ssam2and disable push access to git.baserock.org for baserock/ repos11:42
ssam2haven't done it because there are higher priority things, and i'd want to check for concensus before doing it11:42
persiaI still don't like having two, but that would make the automation less fragile.11:42
ssam2one-way replication is a pretty standard thing11:42
ssam2although it could be seen as unnecessary in this case11:43
persiaI think it is necessary, because we're depending on the reliablitity of gitano for provenance reasons.11:43
persiaI'm not convinced it can be gotten right with gerrit, but as I said before, I don't want to move from gerrit until we find something that does do everything right.11:44
straycathi, any reviews of https://gerrit.baserock.org/#/c/959/ and https://gerrit.baserock.org/#/c/960/ would be appreciated11:56
paulsherwoodstraycat: ukfast?11:58
paulsherwood959 merged11:58
paulsherwood960 merged anyway since it had +211:59
straycati live in the uk so i guess i got redirected to a uk server, in any case fetch speed isn't that important12:01
straycatthanks for merging12:01
paulsherwood:)12:02
pedroalvarezwow, Ansible has a chroot backend! That means that we could create a ansible.configure extension that runs playbooks12:03
persianifty.12:04
pedroalvarezbut I think this will break cross-deployments (deployments of systems of different architectures)12:05
persia:(12:09
persiaMaybe something can be done with qemu-static?12:09
pedroalvarezSeriously, creating/modifying users, groups etc in Baserock should be easier12:28
*** cosm has joined #baserock12:33
*** sherm_ has quit IRC12:36
*** sherm_ has joined #baserock12:49
*** fay_ has quit IRC13:04
*** fay_ has joined #baserock13:19
straycathi, wrt https://gerrit.baserock.org/#/c/962/ is it ok for me to run update-ca-certificates on gbo before we merge that?13:37
straycatKinnison, ^13:37
pedroalvarezg.b.o has been updated no long ago. Why would running that be needed?13:40
pedroalvarezOk, just seen your comment13:41
ssam2it might be that we need to update ca-certificates in the reference systems too, then?13:46
ssam2as in, update the ref of it that we include13:46
straycatyes i think so13:50
straycathrm, i guess running that ourselves could break an upgrade? so should be avoided13:52
straycatprobably better to update the sha and upgrade the trove13:52
KinnisonErm update-ca-certificates is just about processing the locally installed stuff isn't it?13:56
ssam2i thought so, yes13:56
straycatokay, i might have made a mistake somewhere then13:58
straycatfeel free to merge it without doing anything in that case13:59
straycatalso while we're here any reviews of https://gerrit.baserock.org/#/c/961/ would be very useful14:01
*** franred has quit IRC14:01
paulsherwood+114:01
straycatthanks14:02
paulsherwoodssam2: is there a way to call sandboxlib with warn=False?14:21
ssam2I think you can just set warn=False14:21
ssam2i'm not a fan of adding options that don't actually work if you choose one of them14:22
paulsherwoodssam2: could you expand on your last comment please? i'm confused14:22
ssam2in YBD right now there's this line: config = executor.degrade_config_for_capabilities(config, warn=True)14:23
ssam2seeing that, you can expect that warn=False works too14:23
ssam2if it doesn't, it shows i'm a dreadful programmer ;)14:23
paulsherwoodah, ok. i approached this from the wrong place. i had assumed that i would need to adjust the actual call14:24
paulsherwoodis the api/calling convention for sandboxlib documented? there's lots of useful background on the readme/homepage but no calling info afaict?14:25
ssam2there are docstrings in https://github.com/CodethinkLabs/sandboxlib/blob/master/sandboxlib/__init__.py14:27
ssam2it'd be nice to set up Sphinx to generate HTML docs14:27
paulsherwoodseems an obvious scope-creep for github pages? :)14:27
ssam2i'm surpised they don't provide that as a service, in fact14:28
ssam2well, they render .rst and .mdwn files nicely when you view them14:28
paulsherwoodexactly14:28
ssam2so they do effectively provide it14:28
ssam2just not the extract-documentation-from-docstrings-in-code thing that Sphinx does14:28
ssam2sandboxlib is simple enough that keeping the docs separate from the code might be OK, though14:29
paulsherwoodssam2: it works, thank you :)14:36
paulsherwoodi've closed https://github.com/CodethinkLabs/sandboxlib/issues/814:36
ssam2cool. although if YBD destroys someone's OS due to a misbehaved chunk sandboxed only using chroot(), you can no longer say "I warned you!"14:37
* Kinnison thinks sam's missing a negative in there somewhere14:38
paulsherwoodssam2: good point. maybe best if i output that warning, more clearly, once, rather than every invocation though14:39
paulsherwoods/i output/ybd outputs/14:40
ssam2yes14:40
* rjek gasps at the state of the coffee machine14:41
rjekw/w14:41
ssam2for now, you could just add 'if sandboxlib.get_executor_for_platform() == sandboxlib.chroot: print "a nice warning"' in ybd.py14:41
ssam2or rather, 'if sandbox.executor == sandboxlib.chroot'14:42
paulsherwooddone ;)14:47
*** mariaderidder has quit IRC15:30
*** jonathanmaw has quit IRC16:11
*** sherm_ has quit IRC16:18
*** petefoth has quit IRC16:48
*** bashrc_ has quit IRC17:00
*** ssam2 has quit IRC17:00
*** flatmush has quit IRC17:09
*** rdale has quit IRC17:09
*** flatmush has joined #baserock17:20
*** flatmush has quit IRC18:28
*** Kinnison has quit IRC19:15
*** Kinnison has joined #baserock19:15
*** lachlanmackenzie has quit IRC19:18
*** zoli__ has joined #baserock19:44
*** zoli___ has joined #baserock19:53
*** zoli__ has quit IRC19:56
*** zoli__ has joined #baserock20:21
*** zoli___ has quit IRC20:24
*** zoli__ has quit IRC20:33
*** zoli__ has joined #baserock20:34
*** zoli__ has quit IRC22:21
*** zoli__ has joined #baserock22:21
*** brlogger has joined #baserock22:23

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