[10:39] <frankban> hi rogpeppe: is statecmd still the preferred package where to put API calls logic? or is it preferred to put everything in apiserver/client/client.go?
[10:39] <rogpeppe> frankban: statecmd is the place to put logic which is called by both cmd/juju and the API server
[10:40] <rogpeppe> frankban: the aim is to eliminate it eventually, when we make the command line tools use the API
[10:41] <frankban> rogpeppe: ack, I guessed so. So, if I need to implement several helper functions called by an API call, is it ok to put them in client.go?
[10:43] <rogpeppe> frankban: yes
[10:43] <frankban> cool
[10:43] <frankban> thank you
[10:43] <rogpeppe> frankban: which API call are you implementing, BTW?
[10:44] <frankban> This week (maybe to day) I'd like to start working on ServiceUpdate. It will include the MinUnits parameters.
[10:44] <frankban> rogpeppe: ^^^
[10:44] <frankban> s/to day/today/
[10:45] <rogpeppe> frankban: what else will it include?
[10:45] <rogpeppe> frankban: i guess i'd naively expect a ServiceSetMinUnits call
[10:45] <frankban> rogpeppe: as discussed with William, some weeks ago, everything: charmulr, settings(yaml), constraints
[10:46] <rogpeppe> frankban: ah, seems reasonable
[10:46] <frankban> rogpeppe: he mentioned that an API like this could help in the process of making this kind of calls transactional, IIRC
[10:48] <frankban> rogpeppe: so, maybe a good strategy could be implementing some helper functions that will be call by the new ServiceUpdate and by the other API call that share the same functionality (e.g.ServiceSetCharm). does it make sense?
[10:49] <rogpeppe> frankban: that seems reasonable. presumably this new call is going to overlap in functionality with some existing calls.
[10:49] <frankban> s/will be call/will be called/: writing today is particularly difficult...
[10:50] <frankban> rogpeppe: yes, as I said, ServiceSetCharm is one example, and others are ServiceSet/ServiceSetYAML
[10:51] <rogpeppe> frankban: is the aim that the GUI actually changes several of these things in the same call?
[10:51] <frankban> rogpeppe: the only difference is that in ServiceUpdate every parameter is optional. This seems easily achievable for strings and structs. For ints, perhaps pointers can help.
[10:52] <frankban> rogpeppe: I am not aware the GUI needs this (it only needs a way for setting MinUnits atm), but in the future hopefully can take advantage of this new unified call. 
[10:52] <rogpeppe> frankban: yeah, i'm just wondering what thing we're aiming to making transactional. if we never want to change a service's charm URL in the same transaction as setting its constraints, i'm not really sure i see the point of conflating the two calls
[10:57] <frankban> rogpeppe: maybe I am wrong, and I misunderstood what William told me some weeks ago: maybe the goal was to prevent partial successes?. anyway, I am pretty sure that he requested the new ServiceUpdate API to be implemented, and this will allow the GUI to set the MinUnits for a service. I believe I will also need to take care of including service.MinUnits to the megawatcher for the service.
[11:00] <rogpeppe> frankban: ok. i'm concerned about the fact that this means we'll want to deprecate the old calls, and i'm really wanting to keep the client api as stable as possible, but i guess that's just a hit we'll have to take
[11:03] <frankban> rogpeppe: yeah. I guess ServiceSetMinUnits would work for us as well, but I was clearly pointed to this other direction
[11:04] <rogpeppe> frankban: yeah, i'd like to bring this up with william but he's incommunicado. you'd better go with what he said.
[11:04] <frankban> rogpeppe: cool, thank you
[12:52] <gary_poster> antdillon, hi
[12:52] <antdillon> Hi gary_poster 
[12:53] <gary_poster> antdillon, I have a few options for you
[12:53] <gary_poster> 1) style the zoom widget
[12:54] <gary_poster> 2) fix the export widget in the inspector (it doesn't upgrade; haven't checked out why)
[12:55] <antdillon> gary_poster, Sure I'll grab the zoom widget. That shouldnt take me long
[12:56] <gary_poster> 3) put the purple circle in for landscape issues in the inspector
[12:56]  * frankban biab
[12:56] <rick_h> gary_poster: do you know who can answer gnuoy's question on the deployer in #juju? 
[12:56] <rick_h> gary_poster: http://paste.mitechie.com/raw/993/
[12:57] <gary_poster> (luca can tell you where that is :-) )
[12:57] <gary_poster> 4) thanks rick_h 
[12:57] <gary_poster> heh that was not 4
[12:57] <gary_poster> rick_h, on freenode #juju?
[12:57] <rick_h> gary_poster: yes, that's where he asked it. I figure it's an IS thing? 
[12:58] <gary_poster> thanks rick_h replied there
[12:59] <rick_h> gary_poster: yep, thanks. 
[13:02] <gary_poster> antdillon, #4 was "see if something else is broken and tell me about it and fix it if no one else is working on it" ;-)
[13:03] <antdillon> gary_poster, Awesome, I'll work through them in that order I think, seems to make sense
[13:03] <gary_poster> cool antdillon thanks
[13:39] <sinzui> jujugui webops are urging us to provide nagios integration to the juju-gui charm. Does someone have time to review my branch? https://code.launchpad.net/~sinzui/charms/precise/juju-gui/nagios/+merge/177588
[13:39] <benji> sinzui: I'll be glad to take a look.
[13:40] <benji> sinzui: in way of quid pro quo I would like to know who the best person to talk about the jujucharms.com charm would be
[13:41] <sinzui> benji, I and abentley are the most familiar in both authorship and deployments
[13:42] <benji> sinzui: I was thinking more on the ops side.  I need to verify (through some means that I have yet to devise) that the charm ops used for the GUI includes our latest changes (cache headers)
[13:45] <teknico> benji: fyi, when I run "apache2ctl configtest" on a charm deployment, I get this output: http://pastebin.ubuntu.com/5950894/
[13:46] <benji> teknico: that's something I should look into; thanks!
[13:47] <benji> teknico: hmm, when I run it on a box with a deployed GUI charm I get "Syntax OK"
[13:47] <teknico> benji: also, the config/apache-site.template file uses both tabs and spaces for indentation :-)
[13:47] <benji> ok, now THAT is a problem 
[13:47] <sinzui> benji, gunoy and thedac are best. gnunoy is online now in fact.
[13:47] <teknico> :-)
[13:47] <benji> sinzui: cool, thanks
[13:47] <teknico> benji: there might be some problem with my deployment then
[13:48] <benji> teknico: I added a "a2emod" call to the dependencies setup step that should have added mod_headers
[13:49] <teknico> benji: I have not seen that, are your cache changes in trunk already?
[13:50] <benji> teknico: yep: http://bazaar.launchpad.net/~juju-gui-charmers/charms/precise/juju-gui/trunk/revision/71
[13:50] <frankban> it's interesting that in Italy we use "quid pro quo" in a different way, meaning "misunderstanding". and we express the "exchange of services" meaning using another latin sentence: "do ut des"
[13:52] <benji> frankban: interesting; it seems that the "substituting one thing for the other" (i.e., misunderstanding) meaning is closer to the original usage
[13:58] <abentley> benji: I am thinking of going back to versioning, but before I do, is there anything left from Friday that we still need to put to bed?
[13:59] <benji> abentley: not that I am aware of.  You might ask Gary if he needs anything for the demo [it may be done now, for all I know]
[14:00] <abentley> gary_poster: Is there anything you need for the demo (or is it over)?
[14:01] <gary_poster> abentley, it is over.  server was down again but it was ok.  people still very happy with progress.
[14:02] <abentley> gary_poster: That's a shame.  It's surprising it's so tempermental-- it's deployed exactly the same way as staging.
[14:03] <abentley> gary_poster: I'll rip it down now unless you want to keep it around for some reason.  I expect we'll have staging updated to work similarly soon.
[14:04] <gary_poster> jujugui, orangesquad, busy now but wanted to say generally that people (including MS ;-) were very pleased with the demo.  Juju group has a meeting this evening so may get more details this evening, but so far so very good.  Thank you!
[14:04] <sinzui> frankban, :) In British English we use "to table" to mean lets talk now, but in America English the phrase means talk later.
[14:04] <abentley> gary_poster: Great to hear.
[14:04] <rick_h> gary_poster: woot!
[14:04] <gary_poster> abentley, yeah, rip it down.  sorry it didn't work out but glad for progress nonetheless
[14:07] <frankban> :-)
[14:20] <benji> sinzui: where is the best place to contact gunoy, I don't see him anywhere I normally /join
[14:21] <rick_h> benji: he was in #juju eariler today
[14:21] <rick_h> earlier ugh
[14:25] <sinzui> benji, #webops
[14:52] <hatch> morning all
[14:53] <antdillon> Hi, is there a way to clean the css its appending the new styles?
[14:53] <hatch> say what?
[14:53] <rick_h> lol
[14:54] <antdillon> When I refresh a local juju gui it add's the combined css from the less to the end of the existing css
[14:55] <antdillon> Does that not happen for everybody else?
[14:55] <hatch> so you're getting *.less content at the end of the css file?
[14:55] <rick_h> antdillon: I think you need to go a little deeper. 
[14:56] <rick_h> antdillon: yes, normal build is to process the less styles and build a combined stylesheet out of them
[14:56] <rick_h> antdillon: I don't think we have an 'unprocessed' version of it in single files available right now
[14:57] <antdillon> hatch, rick_h Not less content, this combining the css and adding it to the end of the css file
[14:58] <antdillon> rick_h, My juju-gui.css file is 27000+ line long at the moment
[14:58] <rick_h> antdillon: hmm, 16,221 here
[14:58] <antdillon> I guess just deleting the contents of juju-gui.css and re-running would do the trick
[14:59] <rick_h> antdillon: right, I understand what you're saying, but we're not following what you're asking. You want to rebuild the file? You want to not combine the file?
[14:59] <hatch> antdillon: maybe run `make clean`
[14:59] <hatch> rick_h: I 'think' that it's concatenating the contents to the file instead of remaking it
[14:59] <hatch> is what he is saying
[15:00] <antdillon> rick_h, Yes sorry I want to combine but it retaining the css from before the change
[15:00] <rick_h> hatch: oh hmmm, antdillon what commands are you running to 'rebuild' the css?
[15:00] <antdillon> hatch, Yes appending to new combined styles to the end
[15:00] <antdillon> rick_h, Just refreshing the localhost
[15:01] <rick_h> antdillon: if yuo wanted to try to keep your new stuff isolated I'd add a css file to index.html and hack on it. Then combine it into the right places in the less world. Any less file changes should auto-rebuild the css file for you. 
[15:01] <antdillon> hatch, So every refresh is adding over a 1k of lines to the css
[15:01] <rick_h> antdillon: want to chat on a hangout? maybe easier to follow/suggest something?
[15:01] <hatch> well that's not possible hehe
[15:01] <hatch> so something is broken :)
[15:02] <antdillon> rick_h, Oh ok, I can split it off to its own. Just wondered if instead of concatenating the new combined styles it could replace
[15:03] <rick_h> antdillon: well that's why I ask what commands you're running? It should auto rebuild on its own as you edit the existing files.
[15:03] <hatch> i'm still curious as to how refreshing the browser is adding to the css file
[15:03] <rick_h> antdillon: if you manually run some less command, it might act strange
[15:03] <rick_h> hatch: yea, it can't be. So there's some step we're missing. Why I suggest a hangout at this point
[15:03] <rick_h> to see what's up
[15:04] <antdillon> rick_h, hatch No commands at all, I had a few changes to stylesheet.less and refreshed a few times and its combining the styles on each refresh but not replacing the content in juju-gui.css but appending it to the end
[15:05] <antdillon> rick_h, So If I give something some styles and then remove them it keeps that on earlier lines.
[15:05] <antdillon> rick_h, I have to set them back to defaults
[15:06] <hatch> antdillon: and if you run `make clean` and then `make devel` again?
[15:06] <rick_h> antdillon: ok, I see it. It's broken. 
[15:06] <rick_h> antdillon: it should not be doing that. 
[15:06] <rick_h> antdillon: so you found a bug :)
[15:07] <antdillon> rick_h, Woop lol
[15:07] <antdillon> hatch, Running them now
[15:07] <rick_h> filing and will see what we can do about it. 
[15:07] <rick_h> antdillon: we'll get you a short command to use for the moment. Sec
[15:07] <antdillon> rick_h, Cool, are you not moving over to SASS?
[15:08] <rick_h> antdillon: trying to, but not today
[15:08] <antdillon> rick_h, COol
[15:09] <antdillon> rick_h, Oh by the way not sure if you've seen emails yet but I set up a frontend mailing list for canonical if your interested
[15:09] <antdillon> hatch, Worked a treat, thanks
[15:12] <hatch> antdillon: np - that's the goto 'fixall' :D there is also `make clean-all` which will also remove all of the node modules
[15:12] <hatch> antdillon: I joined your mailing list...do I get a grab bag or something now? :)
[15:15] <rick_h> antdillon: this will 'reset' faster than a make clean will. 
[15:15] <rick_h> rm build-shared/juju-ui/assets/juju-gui.css build-shared/juju-ui/templates.js  && make build-shared/juju-ui/templates.js
[15:15] <rick_h> antdillon: just keep that in your command history for doing a 'reset' on the css and should help. Filing a bug now
[15:16] <hatch> rick_h: do we know how to repro the bug?
[15:16] <rick_h> hatch: yea
[15:16] <hatch> oh ok cool
[15:16] <rick_h> hatch: #1208503
[15:16] <_mup_> Bug #1208503: juju-gui.css rebuilds over itself improperly <juju-gui:Triaged> <https://launchpad.net/bugs/1208503>
[15:17] <hatch> rick_h: so you're telling me this has been happening for almost a year and noone has noticed?
[15:17] <rick_h> hatch: looks like the less build command in templates.js is doing a "appendFileSync" which I assume means it's appending vs rebuilding the file
[15:17] <rick_h> hatch: well think about it. cascading CSS
[15:17] <hatch> lol
[15:18] <rick_h> it appends the new rules, which cascade over the old ones
[15:18] <rick_h> so it didn't 'effect' the layout/etc
[15:18] <hatch> that would be 'cascading cascading style sheet'
[15:18] <rick_h> n + 1
[15:18] <hatch> haha
[15:18] <rick_h> hatch: so since it's in the node stuff I'll assign it to you :P
[15:18] <hatch> sure thing :)
[15:19] <antdillon> hatch, Thanks, yes its in the post!
[15:19] <antdillon> rick_h, Thanks
[15:21] <hatch> well all of the stump grinders in the city are rented so I gota get back to digging this damn thing out :/
[15:21] <rick_h> grinder!
[15:21] <Makyo> jujugui meeting in 10 according to calendar, kanban now
[15:21] <rick_h> <3 those things. 
[15:21] <hatch> oh standing up is in a couple minutes
[15:21] <Makyo> 9 now.
[15:21] <rick_h> yep
[15:21] <hatch> maybe I'll hang around then
[15:29] <Makyo> jujugui call in 1
[15:29] <bcsaller> how did I miss this change? :-(
[15:30] <rick_h> trying to get in but no load on the page :(
[15:55] <teknico> does anyone fancy reviewing some more charm refactoring? :-) https://codereview.appspot.com/12467044
[15:58] <gary_poster> jujugui, orangesquad, hi.  I'm afraid I have a meeting with MS in conflict.  However...
[15:59] <gary_poster> I just forwarded an email to peeps
[16:00] <gary_poster> Could you start the next hour by reading its content? :-)
[16:00] <rick_h> gary_poster: so hangout is not taking place now?
[16:00] <gary_poster> rick_h, not with me, no
[16:23] <rick_h> hah "It's better to have something like 'the number of commits in the last month'"
[16:25] <rick_h> so what were they acutally using in this? Was it some sort of mockup?
[16:28] <gary_poster> rick_h, in part
[16:29] <rick_h> gary_poster: ok, I guess it says it used the 'prototype' in two states but assumed it was a juju-gui install
[16:30] <gary_poster> y
[16:30] <gary_poster> used juju-gui for part
[16:30] <Makyo> Some of these run counter to goals expressed to us as a team in Oakland.  Would be interested to discuss after IoM. 
[16:31]  * gary_poster has not read yet; will do so tonight.  We can hopefully talk about it this week, even.
[16:31] <rick_h> gary_poster: gotcha. Yea some of these confuse me and I assume it's how the mockup version worked. Not sure how they 'changed the tab' 2/3 way down. The back button works for going back to the charm after clicking a related one, hooks is gone, ti's source. 
[16:31] <rick_h> very cool to read though. Some good info in here
[16:32] <rick_h> always fun to have testing back up things that have been said/brought up. 
[16:33] <gary_poster> cool
[16:34] <gary_poster> jujugui, orangesquad, please be welcome to send comments to the mail
[16:35] <abentley> gary_poster: I don't know which mail you mean.  After you mentioned it, I looked at the juju-gui ML and didn't see anything.
[16:38] <rick_h> abentley: PM
[16:40] <gary_poster> abentley, juju-gui-peeps
[16:40] <gary_poster> abentley, private
[17:28]  * benji reboots to do a little post-lunch RAM upgrade.
[18:36] <benji> abentley: I'll be ready to join up in a couple of minutes if I'm not too late to the versioning party
[18:53] <abentley> benji: Okay, let's hang out whenever you're ready.
[18:59] <benji> abentley: guichat is open
[21:16] <hatch> hey all
[23:08] <huwshimi> Morning
[23:53] <hatch> morning huwshimi
[23:53] <hatch> thanks for the update on the build script
[23:54] <huwshimi> hatch: Hey
[23:55] <huwshimi> hatch: I was going to push a branch up for that today
[23:55] <huwshimi> hatch: It's a tiny fix