[03:32] Makyo: I was thinking the juju terminal console - wouldn't he be able to get it from there? Improv should be dumping that sort of stuff no? === free`` is now known as free` [08:20] * mariusko thinks he hits bug #1149410 again... [08:20] <_mup_> Bug #1149410: Stuck on "Trying to connect to the Juju environment" < https://launchpad.net/bugs/1149410 > [08:21] mariusko: what provider are you using? [08:22] mariusko: and what version of the GUI? [08:25] frankban: AWS [08:26] I changed back to a stable version. That is juju-api-branch: lp:~hazmat/juju/rapi-rollup and juju-gui-source: stable [08:26] Worked for some time. I also sometimes have more success using IP address that DNS name [08:53] mariusko: you switched back to stable using "juju set"? [08:57] frankban: nope, it was a new installation [08:57] new deployment [09:02] mariusko: any javascript console errors? [09:09] frankban: "Application Cache Error event: Master entry fetch failed (-1)" [09:10] mariusko: humm.. are you using chrome/chromium? [09:10] frankban: chromium. Isn't FF still not supported? [09:17] mariusko: try to delete the appcache: go to chrome://appcache-internals/ and remove the entry corresponding to your ec2 machine === andrewdeane is now known as awd [10:01] frankban: same problem with FF. Python process on the machine is using a lot of CPU [10:02] but it has 1,7 GB RAM [10:56] "WebSocket connection to 'wss://x.x.x.x/ws' failed: WebSocket is closed before the connection is established" [11:02] frankban: debugging further, /ws gives "503 Service Unavailable: No server is available to handle this request." [11:03] I see that there is a haproxy there, and "http://127.0.0.1:8080/" gives nothing [11:06] Uhm, even "/var/lib/juju" is gone on the server now [11:44] mariusko: any error in /var/log/upstart/juju-* [12:25] did something change with the tests? The ones loading json are failing for me as the paths to the test/data/xxx seems to have changed? [12:25] bah, some test is monkeying with the url adding a / to the end in chrome causing the paths to fail [12:29] rick_h_: make test-server works well for me in trunk [12:30] frankban: yea, there's a test half way through that's adding a / after index.html and causes all the relative paths to test data files to be index.html/data/xxxx.json [12:30] frankban: might be an interaction with chrome I've got here [12:31] hmm, does it for me in test-prod as well though so must not be a chrome thing [12:33] nvm, guess it's something in this branch. [12:43] hatch: when you get a chance can you peek at/QA the button fix you noticed yesterday? https://codereview.appspot.com/8363043 [12:44] frankban: killed the instance. Will check next time. New instance working. [12:45] mariusko: ack [13:25] hatch: ping when you're around [13:25] bac: looks like the branch yesterday went ok then? Caught the reviews after EOD. [13:26] rick_h_: yeah, just doing some clean up [13:26] bac: I didn't lead you too far astray when what they meant I hope? [13:26] cool [13:26] thanks for your advice [13:32] hatch, re juju terminal/improv - that was a live install, not improv, but there is a chance the bootstrap node would have it in its logs; I don't know if that all gets logged. [13:36] Also, frankban, last daily for the tablet just about bricked the thing, so I was screwing with that all yesterday, sorry I didn't get more review done; I'll do that first thing. [13:36] Makyo: no problem, and thanks. [13:38] * frankban fighting against allhands [13:55] hatch: do you need a review for "implement unexpose"? [13:57] benji yeah looks like i need one more [13:57] hatch: I'll be glad to. I don't see the URL on the card, do you have it at hand? [13:57] https://codereview.appspot.com/8339044/ [13:57] thank yas [13:58] bac: I'm here [13:58] hey hatch [13:59] thanks for the excellent review. i only had one question about one of your comments. i think my question is in the review now. [13:59] alright lemme check [13:59] first comment at https://codereview.appspot.com/8338043/diff/5001/app/store/endpoints.js#newcode24 [14:00] the endpoints map and subscriptions is the properties I was talking about - if you define them in the initializer then they are local to the instance not on the prototoype [14:00] that's what I mean [14:00] wrt 'following properties' [14:02] so judging from your comments you already did that [14:03] brb [14:04] hatch: cool. please update the review if you think it is landable [14:17] bac: endpoints.js has a syntax error... [14:18] lol how did that get through the linter [14:18] ln 44 [14:20] hatch: it is a style issue but not syntax i think [14:20] fixed [14:21] oh right [14:21] that's what i meant :) [14:22] LGTM'd [14:22] thx [14:22] hatch: i still have an open question about reset as i'm unsure when that is called. [14:23] hatch: should reset call unbind? [14:25] if there aren't any endpoints in there would you still want to listen to those events? [14:26] if reset calls unbind then you'll need to rebind them again no? [14:30] yes [14:31] i'll land as is and pick ben's brain later. [14:35] benji: do you have some notes re missing go watcher attrs? I am filing a bug I will work on after my current handlers/converters refactoring branch is merged [14:36] frankban: I turned the ones I found into cards. [14:39] benji: are they in story 1? I cannot find them in the board [14:39] frankban: I only found two and one is done now. [14:40] perhaps you found a treasure trove of missing data that I missed. [14:43] benji: no, I am just trying to figure out what's missing. [14:53] rick_h_: is huw using OSX and that's why he can't lbox? [14:58] benji: do you have a minute for a quick call? [14:58] hatch: no, he's had a few diff errors. A couple look like LP timeouts. Another was just problem figuring our linting. CSS rules matching xxx_yyy vs xxx-yyy don't actually output a message [14:58] hatch: so he's just not spent time figuring it all out so he's getting a pass [14:59] frankban: sure [14:59] frankban: the normal hangout? [14:59] benji: yes, I am there [15:00] frankban: hold on, the Google hangouts plugin isn't working [15:01] rick_h_: ahh ok [15:05] rick_h_: I'm a little confused as to why we are using images for that add button - I don't see anything there that coudln't be done with a icon font and some css [15:06] hatch: I'm not sure at all. Since we're getting .pngs from UX I'm not sure if it's a time thing to pull it out or if there's something about it. We can bring it up. [15:07] alright thanks - yeah because that's 3 http requests right there which could be one icon font and some css [15:08] I'll approve it but just make a note please :) [15:09] hatch: will add a discussion card to our board [15:09] (Y) [15:11] oh look at that CI broke [15:22] benji: reviewed your branch. [15:22] bac: the go-juju one? [15:22] oh hmm, doesn't look like I can actually do destroy_service until remove_relation is done [15:22] benji: yep [15:23] bac: is your review "binding" (i.e., can I count it as my second review? please say yes, please say yes...) [15:23] guihelp: i had a pre-existing committment at noon today before we changed our meeting time. i'm going to have to miss the meeting today. [15:23] :'( [15:24] benji: sure. i'll be happy to LGTM it when fixed [15:24] bac, alright. Anything about your current work we ought to know? [15:24] Makyo: it is stellar [15:24] :-) [15:24] \o/ [15:24] bac: what I mean is, are we appoved as full reviewers for go-juju? [15:24] Makyo: branch landed. doing misc stuff right now (lp2kanban, reviews, etc) before picking up another item after lunch [15:24] benji: yes, that is my understanding [15:25] bac, sounds good. [15:25] excellent [15:25] Makyo: i'm open to suggestions on what to do next if the crowd has ideas [15:25] benji: the times i've prefaced reviews with "i'm not an official reviewer" the juju-core guys pushed back and said "you're good enough" [15:25] bcsaller: are you in yet? [15:26] k [15:26] benji: why the lack of tests on that branch? [15:27] bac: I'm confused. It has tests, not many, but it doesn't do much and it follows the same pattern as the other branches that do similar things. [15:27] hatch: yes, mostly [15:28] benji: did i miss something? i just saw the one api_test [15:28] i may have missed it due to multiple revisions on the rietveld [15:28] bcsaller: ok I THINK the order I need to do things are... add_relation > remove_relation > destroy_service [15:28] because you can't destroy a service unless it has no relations [15:28] and you need to be able to remove the relations if it has some [15:28] but you can't remove any if you haven't added any [15:29] yay TDD [15:29] :D [15:29] bac: yep, that's probably all there is; for better or worse that is what has been deemed acceptable. [15:29] bcsaller: so what I"m asking is if that order of events makes sense to you [15:29] benji: ok. i expecte to see something in statecmd. but, as you say, it doesn't do much. [15:30] "expecte" -- it's like i'm speaking french now [15:31] expecté [15:31] hatch: that makes sense to me. [15:31] word [15:31] these look like large tasks [15:32] hatch: they kind of are, my stuff turned out to be bigger than I thought. W/o relying on the validation juju does (for improv) there is quite a bit to check on any mutation [15:33] hatch: I need to finish the ones I'm doing but I'm open to pairing on it later if you want that [15:33] well last night I was dabbling in some C# so I'm glad to be back in javascript...:) [15:34] c# has 4 different 'length' methods depending what type of array you have....cuz that makes sense [15:35] :P [15:38] although the most irritating is the fact that methods AND classes have an uppercase first letter so you have no idea what can be instantiated by looking at the name [15:54] bac: I have fixed the errors you pointed out and re-proposed the branch. (https://codereview.appspot.com/8366043) [15:54] guihelp: the javascript console (trunk) shows this error when double clicking a service: http://pastebin.ubuntu.com/5677068/ [15:55] that looks like an error caused by bac's changes [15:55] indeed [15:55] bac are you still around? [15:56] oh dear gjslint, why do you explode rather than telling what it is that you actually don't like? [15:56] frankban: maybe create a ticket and assign to bac so he can fix when he returns? [15:56] it's not good for your health, you know [15:57] hatch: will do [15:57] jujugui guichat in 2 === deryck is now known as deryck[lunch] [16:01] bcsaller, rick_h_, luca____ ping [16:10] https://codereview.appspot.com/8345046 [16:14] teknico: I'll take one [16:14] hatch, thanks! [16:18] done [16:21] hatch, thanks! (remember to tag yourself on the card, actually *before* you make a review :-) ) [16:21] speaking of which: benji, did you say you already got two reviews for your branch? I see no tags on its card [16:22] tagged [16:22] (playing card tag police here :-) ) [16:22] hatch, thanks again :-) [16:23] teknico: only one of the reviewers is on the gui team, the juju-core guys don't use that technique. [16:23] my self-imposed rule is that only reviews for projects completely "inside" the gui team use tags to claim reviews [16:24] benji, but bac said that we can be authoritative reviewers on juju-core too, so why not? [16:24] I mean, we could add tags for reviews from the juju-core team in their place [16:24] if we can get them to participate, then it would be great; partial participation seems worse than non-participation though [16:25] we could do that [16:27] Makyo, thank you too (I added your tag :-) ), landing! [16:27] teknico, good catch, thanks [16:51] that moment when you realize you just deleted the wrong folder... [16:53] yeah, that one :-) [17:13] in python `for pair in endpoint_pairs[1:]:` what does the : inside the [] mean? === deryck[lunch] is now known as deryck [17:19] hatch: that : makes it a slice, so [1, 2, 3, 4][1:] == [2, 3, 4] [17:19] oy [17:19] ahh ok [17:19] thanks [17:19] the first index goes before and a second index goes after, if the first is left off it means 0, if the last is left off it means -1 (last) [17:20] bac: https://bugs.launchpad.net/juju-gui/+bug/1164589 [17:20] <_mup_> Bug #1164589: addServiceToEndpointsMap error < https://launchpad.net/bugs/1164589 > [17:20] * bac looking [17:24] bac: just updated with a new comment on that ticket [17:27] bac: that error is actually blocking me so I can help if needed [17:27] hatch: first one was easy. looking into the other [17:28] oh ok excellent - be sure to now write a test to cover that error :) [17:32] hatch: yes dear [17:32] lol [17:32] hatch: after the fix for the first i cannot replicate the add relation issue [17:32] egggggggcelent [17:34] is there a special procedure for staff purchases from shop.canonical.com? [18:11] hatch: re: purchases - no. if they are still doing the discount code, you should've gotten it when joining. === BradCrittenden is now known as bac [18:18] hatch: proposing fix now [18:21] kewl [18:21] thanks [18:26] hehe fix-goof [18:34] bac: a 4 character fix? hehe [18:34] i think it is eight. i had to delete 'null' before adding 'this' === BradCrittenden is now known as bac [18:35] oh lol - charmStore needs to be destroyed no? [18:35] in your test [18:38] no codereview url has come by so that's why I mentioned in here :) [18:39] https://codereview.appspot.com/8364044 [18:39] hatch: could you qa this branch to see if you can cause the add relation failure? [18:41] yep checking it out now [18:42] jcsackett: hatch if you guys get a few min please? https://codereview.appspot.com/8356044 it's a step one so you can qa it that it works, but it's fugly and only works via direct url calls (can't click into the view) [18:43] benji, rick_h_ : could one of you do a quick review on https://codereview.appspot.com/8364044 [18:43] bac: sure thing, loading up [18:43] thx [18:43] bac, QA'd and OK [18:43] missed it [18:43] thanks hatch [18:45] rick_h_: I still don't see a position absolute in this branch :P [18:45] hatch: nope :) [18:45] hatch: honestly, last thing we'll do. This way we know any bugs are our own and not interactions [18:45] well, maybe not last... [18:46] bac: if you get a second, I could use a +1 on the review from earlier [18:48] benji: done [18:48] bac done [18:49] rick_h_: so no more issues with the routing code? [18:49] thanks rick_h_ [18:49] hatch: :/ only issue is the double/triple call which I'm ignoring for now [18:49] hatch: goal is to try to get the basics setup for huw to go through and be able to apply UX love to things [18:49] oh yeah but that was an issue long before the routing stuff [18:49] hatch: ah, well I wasn't really involved before so meh [18:50] neither was I :P [18:50] hatch: I do forsee one issue I need to work through, but haven't looked awt the problem much yet [18:50] I really wish I had a way to generate urls in handlebars based on something with some brains on current state [18:50] because if I do the url routing in my views by catching click events I don't get dispatch [18:51] but in the templates there's no reall app state to say "oh, you're in /fullscreen so all links should stay in /fullscreen land [18:57] rick_h_: create a helper which references the generator in the router code [18:57] I generate urls that way in the service views [18:57] hatch: cool, will look there then. [18:57] but instead of using a helper I do it before passing it in [18:58] oh cool, yea dirty thing would be for the master view to set an magic ATTR that's 'whoami' or something but it's nasty/dirty [18:58] serviceRemoteUri: this.get('nsRouter').url({ gui: '/service/'}) [18:58] for example [18:58] so that says, change the url for the gui namespace to /service/ [18:59] well....build me a url with the gui namespace with /service/ [19:00] hmm, yea the big thing will be the CharmToken widget, when clicked on, will need to open /bws/XXXX/precise/charmid to load the details [19:01] in sidebar, that's this expanded sidebar version, in fullscreen, it's the fullscreen details view [19:01] where sidebar/fullscreen is XXXXX [19:01] so need to get something into the CharmToken Widget to be smart about generating that url/click [19:01] yup [19:06] thanks bac [19:10] bac: so in testnig it out YUIDoc doesn't care about the lack of return. The methods don't return anything and rather than add more noise with the whole mutates only mess I just leave it out. [19:10] at least make code-doc and make view-code-doc don't complain [19:10] rick_h_: that is great news and should be shouted from some mountain top or roof [19:15] * hatch is back on track [19:15] thanks for the patch bac [19:20] bcsaller: have a second to explain to me how relationships work from the higher level? I see the code but I'm wondering how they function inside each charm....or if this is documented somewhere I could read that instead :) [19:21] hatch, one min [19:24] sure [19:24] hatch: in hangout [19:44] statik signed on at Mar 10, 2013 20:37:57 and has been idle for 594hr 59min 51s [19:44] i guess i'll try email [19:47] lol [19:50] bcsaller: so that I am sure I get this right [19:50] relationships are valid if [19:50] they are the same type... [19:50] one is client, one is server [19:50] both are 'peer' [19:52] I'm not sure how peer to peer differs from 'both same type' but they are checked for separately [19:52] hatch: at a high level yes, there is also a notion of subordinate scoped relations which have the same basic rules but generally use an implicitly provided relation, 'juju-info' [19:53] hatch: with client-server, one is client, the other is server (roles) and they have the same interface type [19:53] hatch: with peer they are peers, they all have the same role [19:53] ahh ok [19:54] hatch: the special behavior around peer relations is that they are automatically joined on deploy [19:55] hatch: there is no explicit step needed because they are 'related to self' [19:55] hatch: you can see that at the bottom of juju/control/deploy.py [19:56] (which I don't know if we implement in our version) [19:56] doesn't look like it [19:56] nope [19:56] hatch: did you do anything to fix jenkins or was it not a real failure? [19:57] * bac trying to get all set up with vpn, etc [19:57] there was a lot of work done to do all this in python to throw it all away and redo in go... [19:57] bac: I think my land came alnog after yours and forced a re-run? [19:57] bac: I just left it, I'll do it towards the EOD [19:57] bac: but guessing as I thuoght the failure on yours was mine [19:57] rick_h_: no it complets and then does it again, it doesn't stop [19:58] see, what do I know about how it's setup? nadda :) [19:58] hatch: so we assume the failure was spurious/canonistack [19:58] hatch: we really need to fix our use-cases for this thing, we don't want to build all the juju behavior in JS I think, just get the env running [19:58] bac: I'm going to yes - there is no reason any of our changes have caused that failure [19:59] but it still takes alot to get just as far as we need [19:59] bcsaller: yeah - how I understood it is that we want it to 'look' like it's working as expected [20:00] so I'm sure we can forego some of the more complex checks [20:00] but I am pretty sure we want to check to make sure the relation stuff works properly [20:01] hatch: yeah, its one of the main things we draw [20:10] bcsaller: mind taking a look in topology/relation.js and let me know if I am seeing this wrong...It looks like we only support server and client connections in the GUI? 625-631 and 659-665 [20:19] hatch: peer relations are not ambiguous which is what that method is checking [20:20] hatch: we dont' draw peer relations though as they would look like an arrow pointing back to self or a ring or something. [20:26] well sure but that's the only method which actually calls add_relation [20:26] so right now the gui doesn't actually check the charm to see if it's a server or client [20:27] it just says "you're a client, and you're a server" [20:27] hatch: i'm trying to setup my vpn to the qa lab. you know the drill, right? [20:28] bac: umm mine was a little different because I'm using 12.10 [20:28] so I coudln't follow the guide [20:28] hatch: that's what i'm using [20:28] oh ok, lemme boot up my laptop [20:29] hatch: well, real, quick, once you have network-manager set up, how do you get it to open the vpn connection? [20:29] click the network symbol in the top right [20:29] if i double-click on it, then it just opens the edit configuration dialog [20:30] top right of what? [20:30] nope single click on the thing that looks like a cell phone signal strength (with my icon set) then go down to 'VPN Connections' [20:30] in the very top right of the screen [20:30] ok, i was trying w/in the nm [20:30] then that will open up another menu that has whatever you called the vpn connection [20:31] do you see it? [20:32] honestly my configuration is so messed up I don't think I'll ever be able to upgrade to 13.04 without something exploding :D [20:32] this was also installed with wubi 3 years ago and without wubi in 13.04 I'm not sure I'll actually be able to upgrade anyways [20:33] hatch: cool, i'm in [20:35] right on [20:35] plz don't delete our jenkins configuration :) [20:45] LG Optimus G at my cell provider is $0 on a 3 year contract....or $499 on a two year...lol [20:49] bcsaller: I need the mysql equiv to test/data/wordpress-charmdata.json any idea how I would generate that? [20:51] hatch: http://jujucharms.com/charms/precise/mysql has everything [20:52] hatch: under the repo link you can find the interfaces in metadata.yaml (same as for any charm) but they are also spelled out on the website [20:53] ok so that json file was manually created? [20:57] must be, I can't find any file which has all of this information in it [20:58] hatch: what file are you looking for? [20:58] well I need to deploy mysql along side wordpress in the in memory environment tests [20:58] so I am guessing I need the mysql version of test/data/wordpress-charmdata.json [20:59] k, let me look what api endpoint that is on charmworld [21:00] hatch: http://jujucharms.com/charms/precise/mysql/json [21:00] oh lol it was that easy eh [21:00] :) [21:01] hatch: and that's called from the model.load() using the 'charmstore' store [21:11] kewl [21:12] pro...gress [22:12] bcsaller: reviewing [22:12] hatch: thanks :) [22:40] review done