/srv/irclogs.ubuntu.com/2014/08/25/#juju.txt

=== uru-afk is now known as urulama
=== CyberJacob|Away is now known as CyberJacob
=== balloons_ is now known as balloons
=== HankM00dy is now known as thehe
=== carlas is now known as carlasouzza
=== carlasouzza is now known as carlasouza
=== psivaa_ is now known as psivaa
tvansteenburghstub, cory_fu: would appreciate your thoughts on my latest comment here: https://code.launchpad.net/~tvansteenburgh/charm-helpers/config-implicit-save/+merge/22467814:42
stubtvansteenburgh: I could live with it, but I still think it would be better to hook it into the @hook decorator and save on success.14:51
stubtvansteenburgh: With implicit save, you won't see what has changed since the last hook but what has changed since a previous piece of code changed a value. It is less useful.14:52
tvansteenburghstub: true, i hadn't thought of that14:53
tvansteenburghstub: i'm coming around to the idea of saving in decorator. will just have to document that you don't get the auto-save if you're not using the decorators14:54
stubtvansteenburgh: yes. I think that is better than magic (you could have both modes, with save-on-success being used if a @hook decorator has been entered and save-on-change being used otherwise, but that is probably the worst of all worlds ;) )14:55
tvansteenburghstub: yeah, i think covering the (extremely) common case is sufficient14:56
tvansteenburghstub: thanks for your input14:57
=== scuttle|afk is now known as scuttlemonkey
cory_futvansteenburgh: Sorry, stepped away for a minute.  The problem with "save on success" is that not every charm uses the @hook decorator and so what constitutes "success" varies.  Also, since you can't change config values from within a charm, saving on __del__ is no different (and seems riskier) than saving immediately after reading the previous values.15:00
tvansteenburghcory_fu: i think you're missing the "write your own custom key/val data to Config" aspect15:02
tvansteenburghcory_fu: you can write whatever you want to the object and it'll be saved and available in the next hook15:04
=== CyberJacob is now known as CyberJacob|Away
cory_futvansteenburgh: Hrm.  While that does seem useful, I worry that it implies that you can change the value of config options as visible outside the charm.  I wonder if "store arbitrary data" shouldn't be separated from "easily and detect changes in config"15:06
tvansteenburghmeh15:07
cory_fu:)15:07
cory_futvansteenburgh: stub's concern also raises issues with idempotency when re-executing config-changed hooks during `juju resolved --retry`.15:12
tvansteenburghcory_fu: be more specific?15:13
cory_futvansteenburgh: Just that, in that corner case, running `resolved --retry` may not get you the same behavior because it will detect a different set of previous config values15:14
tvansteenburghcory_fu: yeah, i agree15:15
cory_fu"save on success" would be awesome, if it could be consistent15:15
stubcory_fu: maybe Config and ArbitraryData should be separate, but right now they are conflated.15:15
tvansteenburghcory_fu: but i already agreed to do the save() in @hook, so that covers that concern right?15:15
stubcory_fu: It isn't too late to change that though - the benefit of hiding documentation in docstrings.15:16
tvansteenburghstub: i don't follow re docs?15:17
cory_futvansteenburgh: What about people who don't use @hook and instead use something superior like the services framework?  ;)15:17
stubtvansteenburgh: I don't think many people are using this feature, so we could get away with changing behaviour.15:17
tvansteenburghcory_fu: you can call save() explicitly, or work it into your framework15:17
stubOr better yet, move your framework to charm-helpers so we can all benefit :-P15:18
tvansteenburghcory_fu: i'm not sure what you're arguing for at this point, is there a better place to do the save?15:18
cory_fustub: The docstrings are published online15:18
stubdoh15:18
cory_fustub: Also, the services framework is in charmhelpers now  =D15:18
cory_futvansteenburgh: I can't really think of one, no.  I'm just pointing out that the concept of "on success" isn't universal.  But I guess covering some bases is better than covering none15:19
cory_fustub: Actually it's the second result for "charmhelpers" on Google, even.  http://pythonhosted.org//charmhelpers/15:19
stubSo save will need to be hooked into two places then, since both mechanisms are supported.15:20
cory_fuAww.  My doc improvements haven't been merged & published in those.  :(15:20
tvansteenburghcory_fu, is there a good place to hook it into the services framework?15:21
tvansteenburghcory_fu, have your new docs been merged yet? if so i can push a new docs release15:21
cory_fulazyPower: failed me15:22
* stub guesses reconfigure_services in base.py15:22
cory_futvansteenburgh, stub: Yeah, or manage.15:23
stubyer, manage15:23
cory_futvansteenburgh: Do you have merge powers in charmhelpers?  I'm sad https://code.launchpad.net/~johnsca/charm-helpers/services-docs/+merge/231235 hasn't been merged yet, despite being (practically) nothing but a doc improvement15:26
tvansteenburghcory_fu: sorry, i'm not in Charm Helpers Maintainers15:27
cory_futvansteenburgh: I guess I should bug marcoceppi instead, then.  :)15:29
cory_fuI am sad that I don't get automatically notified of updates to MPs that I've commented on15:34
lazyPowercory_fu: what are you talking about?15:41
lazyPoweri merged that days ago :P15:41
* lazyPower sweeps it under the rug15:41
cory_fu"Charles Butler (lazypower) wrote 2 minutes ago"  ;)15:44
lazyPowermust have been stuck in a queue somewhere15:45
cory_fu:p15:47
cory_fulazyPower: Any chance you also looked at https://code.launchpad.net/~johnsca/charm-helpers/docker/+merge/231226 ?15:48
cory_fuThat one's more substantial, though, so "no" is a fair answer.  :)15:49
lazyPowercory_fu: i skimmed it, but not enough to merge it15:49
cory_fuOk15:49
arosalesmarcoceppi: how did mongodb NY meetup go?16:14
arosalesany video by chance?16:14
lazyPoweras a third party comment re: marcoceppi at NY MONGODB - he came to standup with jazz hands with talk about how to refine our presentation process if your audience is hipsters.16:17
lazyPowerbut ill let him drop the details :)16:17
jcastroJAZZ HANDS!16:18
* arosales anticipates the jazz hand update :-)16:18
cory_futvansteenburgh: Since lazyPower apparently merged my doc changes "ages ago," can you update the published docs?16:24
lazyPowerhaha16:25
tvansteenburghcory_fu: yeah gimme a min16:25
cory_fuThanks.  :)16:25
tvansteenburghcory_fu: updated16:34
cory_fuThanks16:34
marcoceppiarosales: it went really well16:45
* marcoceppi jazz hands16:45
arosalesmarcoceppi: any video of the meetup?16:46
marcoceppiarosales: unfortunately, no16:47
arosalesmarcoceppi: ah ok.16:47
arosalesmarcoceppi: Good to hear it went well, and Jazz Hands were included.16:47
JoshStroblDoc error: https://juju.ubuntu.com/docs/config-vagrant.html has an example for vagrant box add JujuBox that has the URL outside the <pre> tag.16:54
lazyPowerJoshStrobl: the URL being outside of the code display correct?16:59
lazyPowerlooks like how it was rendered, the source looks correct. thanks for pointing this out16:59
JoshStrobllazyPower, yea no problem.17:00
lazyPowerhttps://github.com/juju/docs/pull/14317:01
lazyPowerjcastro, mbruzek, marcoceppi  ^17:01
jcastrodibs!17:01
lazyPowerJoshStrobl: and i stand corrected. vim lied to me. thanks again :)17:01
JoshStrobl\o/17:01
=== roadmr is now known as roadmr_afk
lazyPowerjcastro: you gotta click faster than marco next time ^_^17:02
JoshStrobllazyPower, and that is why I don't use Vim :P17:02
JoshStroblKidding, I just am too lazy to learn its keyboard shortcuts.17:02
lazyPower:set hidden  was the secret sauce17:02
jcastroatom!17:02
lazyPowerthere was a carriage return hiding in there that rendered like the line length broke.17:02
=== jrwren_ is now known as jrwren
JoshStroblalso17:03
JoshStroblhttps://juju.ubuntu.com/docs/authors-charm-quality.html17:03
JoshStroblUpstream Friendly is in the list17:03
JoshStroblshould be outside the list like "Scalable"17:04
lazyPowerWe want that to be the case. Preferred if your charm delivers a mechanism for configureable upgrades, that follow upstream best practices.17:04
lazyPowerOH17:04
lazyPoweryou mean teh bullet point17:04
JoshStroblyea17:04
JoshStroblmuch bullet points, such error17:05
JoshStroblOkay jcastro, here is your chance to be more click happy than marco :P get ready!17:06
lazyPowerhttps://github.com/juju/docs/pull/144 jcastro17:07
jcastro<317:08
JoshStrobl\o/17:08
JoshStrobllazyPower, any idea why there is an unnecessary 4-space "tab" in the <code>juju add-unit -n5 mediawiki</code> section of https://juju.ubuntu.com/docs/charms-scaling.html . I know it is just spacing, just a bit of nitpicking I guess, since there are add-unit calls below that, that aren't spaced that way.17:16
lazyPowerJoshStrobl: probably due to the docs being in a quick rewrite when they were translated to markdown. It was done by a human so it may have some inconsistencies.  If you land a PR we'll review it.17:17
lazyPowerthis is all rendered from src/en/*article-name*.md17:17
JoshStrobllazyPower, sure thing17:18
lazyPower<3 thanks for the contribution/review JoshStrobl17:18
JoshStroblhttps://github.com/juju/docs/pull/14517:19
jcastroman, we are actually making progress on the queue now17:21
jcastroSoon, there will be an end to the dark ages17:21
JoshStrobllazyPower, I'll git pull my fork, if there are a bunch of changes, I'll make them locally, commit them as a group of changes and do a pull request from that so there won't be several pull requests unnecessarily.17:21
lazyPowersounds good JoshStrobl17:32
JoshStroblneeds a review: https://github.com/juju/docs/pull/14617:48
JoshStroblI went through every page real quick and checked for errors. Nothing too thorough since I'm a tad busy :P17:49
=== roadmr_afk is now known as roadmr
=== mwhudson_ is now known as mwhudson
=== JoshStrobl is now known as JoshStrobl[ZZZ]

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!