[03:02] huwshimi: your card ok? I wasn't sure if it's something we could tweak yet until other stuff landed [03:13] rick_h_: It's fine so far. I have the basics working, just need to get all the update/add/remove bits changing the correct elements. [03:14] huwshimi: ok cool [11:28] luca: morning, do you have a link to a ghost relation wireframe? [11:29] Hi frankban, it should look like this: https://docs.google.com/a/canonical.com/file/d/0B7XG_QBXNwY1cjhIZnZXS3pxZG8/edit [11:29] frankban: a grey line with no status indicator in the circle [11:30] luca: thanks, that's what I was looking for [11:33] frankban: great. I can get Spencer to output the assets if you need. [11:39] luca: it would be nice, I just need the circle svg [11:45] frankban: ok, I’ll get him to send it over as soon as he can [11:45] luca: thanks [11:49] luca: abort! I was able to create the svg starting from an existing one [11:49] frankban: ah, nice, no worries [11:50] luca: #e2e2e2 is almost invisible given the current GUI background [11:51] luca: at least with my crappy external monitor. much better using the macbook screen [11:54] frankban: ok, thats the Ubuntu grey but I’ll have a word with Spencer to see if we can get something with more contrast. [12:14] guihelp: I need one review and QA for https://github.com/juju/juju-gui/pull/275 ? Anyone? thanks! [12:14] frankban: sure thing [12:14] rick_h_: thanks [12:36] jcsackett: around? [12:36] bac: ish. [12:37] what's up? [12:37] jcsackett: no rush, when you get in fully would you look at https://codereview.appspot.com/91140045 ? [12:37] bac: sure. [12:39] rick_h_: thanks for the review! [12:39] antdillon: morning, how are you getting along with my email? [12:39] * frankban lunches [13:11] rick_h_, Morning, I am. Getting back up to speed [13:11] antdillon: cool, let me know if you get stuck or have any questions. [13:11] rick_h_, Will do thanks [13:14] bac: lgtm, with a minor question. [13:14] there are no minor questions [13:14] no, wait, there are no dumb questions [13:17] jcsackett: i think what happened here is in an earlier version i was using the 'extra_data' to fill out the expected results, so i wanted it squirreled away separately. [13:17] jcsackett: so i can make the change you request easily and it'll be more readable. [13:18] bac: awesome. [13:28] guihelp how does one troubleshoot bzr connections? [13:28] redir: what's the prob? [13:29] I can't seem to push anything -- it hangs -- and -v doesn't display any more information [13:29] With powerful incantations and blood sacrifice [13:29] That's how [13:29] * redir gets blood bowl and starts chanting [13:30] idea [13:30] found it... [13:30] see, the blood bowl totally works. [13:30] nm [13:31] heh [13:32] stale .ssh ControlPath entry [13:35] bac got a minute to talk about the ngram search interface? [13:42] jujugui added a card to urget, found a big issue while trying to see where we are with machine view in a real lxc env today [13:42] urgent bah [13:42] heads up for folks looking for a new card soon [13:44] rick_h_: I don't have permission to add comments/edit the docs you sent me [13:44] redir: sorry, sec [13:44] np [13:45] redir: updated [13:48] ok, so maybe that bug is my machine [13:48] tx [13:48] jujugui can anyone try to reproduce my bug in urgent please? Now I can't click on a google doc and tried in FF and chrome so maybe my computer hates me [13:48] Checking [13:49] rick_h_: trying [13:49] * rick_h_ feels like his computer is possessed [13:50] rick_h_: I am able move service blocks in LXC, with and without flags [13:50] jcsackett: i did this instead: http://paste.ubuntu.com/7410490/ [13:51] able to move even [13:51] bac: ok, i can see the value of that as a way to isolate expected value. thanks for the changes. [13:51] redir: sorry, didn't see your message. let me get my branch landed then let's talk, whatever [13:51] jcsackett: it is a better test... [13:51] bac: np, ping me when you are free... [13:51] frankban: thanks deleting the card [13:57] redir: want to chat on G+? [13:57] sure [13:58] bac: gimme 5 minutes? [13:58] ok [14:03] bac: ready [14:10] redir: https://plus.google.com/hangouts/_/canonical.com/daily-standup [14:17] morning [14:19] morning [14:20] I can talk today [14:20] yay [14:31] guihelp: I know I've seen this test failure before, but can't remember what it means: "Error: Uncaught SyntaxError: Unexpected token C (http://192.168.33.10:8888/test/index.html:1)" [14:31] That ring a bell for anyone else? [14:31] kadams54 run make lint [14:31] it also seems like it can't serve that file [14:32] if you open that file in chrome devtools you'll see why it's broken [14:32] assuming you're running test server right now [14:36] kadams54: an unmocked api request [14:37] hatch: can you chat to antdillon on watching for changes to the ecs stuff please? [14:37] or Makyo ^ [14:37] sure [14:37] hatch: lint run, still having problems [14:37] hatch, Thanks [14:38] rick_h: it's happening in the after each hook [14:38] kadams54: well that just means that's when the api call responds or something. [14:39] Yeah, I think you're right, looking for API calls… [14:39] kadams54: but the issue is the api response code trying to json decode something that isn't valid json which is what that error is from [14:39] ugh this test suite... [14:39] I can see the error is bubbling out of a json-parse-debug.js [14:40] kadams54: yep [14:41] antdillon ok you can't listen for changes to the changeset [14:41] (that was easy) [14:41] next! [14:41] :P [14:42] hatch, So polling the changeset? [14:42] hatch: boooo, no ATTR to watch? [14:42] antdillon yeah this appears to be an oversight in the design [14:42] :( [14:42] hatch, rick_h_ lol [14:42] should be pretty trivial to fix though [14:43] do you want to tackle that as a separate branch first? [14:43] hatch: antdillon +1 [14:43] hatch, I can try get it in this branch if you know of another example [14:43] It's out of scope of that branch [14:44] it's quite a bit of changes [14:44] basically everything that references the changeSet property needs to instead access an attribute [14:44] OR [14:44] hatch: just need the ECS to fire a change event? [14:44] you hook up to the changeSet property using out object.observe polyfill [14:44] less custom events the better [14:44] ok [14:45] imho [14:45] sure [14:45] so....hmm lemme take a look [14:45] the code is so much cleaner not having to get/set an attribute [14:47] antdillon ever done anything with Object.observe() ? [14:48] hatch, Not really, how hard could it be :/ [14:49] Object.observe(ecs.changeSet, callback) [14:49] not very :P [14:49] hatch, Cool [14:49] haha yeah....you want to look into databinding.js [14:49] no, no he doesn't [14:49] and grep for `observe` and `unobserve` [14:50] but ignore everything else [14:50] it's magic in there [14:50] hatch, Got it [14:51] antdillon basically the pollyfill polls the object :) [14:51] jujugui: nine minutes to call, yada yada. [14:52] antdillon app/assets/javascripts/Object.observe.poly.js [14:52] it's pretty thick, not much to see, but incase you were interested [14:54] we should probably update our polyfill [14:55] hatch: it's a slack card already [14:55] oh cool [14:55] there has been a ton of work done to the Polymer one [14:55] yep [14:56] hatch, Will take a look [14:56] antdillon: you're welcome to join https://plus.google.com/hangouts/_/canonical.com/daily-standup?authuser=1 [14:57] at the top of the hour for stand up [14:57] antdillon so this is just my preference because the code is so much cleaner not using attributes - if you run into any big issues feel free to let me know and we can revisit switching changeSet to an attribute [14:59] jujugui call in 1 [14:59] rick_h_ the blues docs that you shared are not editable fyi [14:59] hatch: k, will look. Thought I gave edit there [14:59] haha, blues... redir [15:00] oh, can edit folder but not docs boo google [15:00] heh, hyper specific ACL [15:00] Makyo: hatch redir call time [15:21] hatch, You in the units in db call? [15:21] I am [15:22] hatch, Let me know when your free for a hand on this observe stuff please [15:26] will do [15:30] antdillon done [15:30] hatch, Think I got it [15:30] nice [15:37] kadams54: want to pair up now? [15:39] Yup [15:39] Working on a PR [15:39] kadams54: cool, setup a hangout when you're ready and linky me [15:40] jujugui marked the unit db list card as critical and top of the lane [15:45] rick_h_: https://plus.google.com/hangouts/_/7acpjs4qa80af83u7shrgk05eg?authuser=2&hl=en [15:51] jcsackett got what you need from that diff img? [16:11] hatch: yup. [16:13] guihelp in running tests I see a ton of "Error: Invalid value for attribute transform="translate(NaN,NaN)" [16:13] this was related to the error I noticed in QA today [16:13] anyone remember seeing or can dupe that? [16:17] Trying now.. [16:18] uhm, so this is confusing, it seems models.js:ServiceUnitList.process_delta already has the code to update both "this" (the units model list itself) and service.get('units')... [16:19] rick_h_: ^^^ so maybe something from the past. anyone remember if in the initial impl we had a top level unit list? [16:19] frankban: so we already have a units list? I didn't see it inspecting the db [16:19] rick_h_, I'm seeing it in test-server. investigating quiickly. [16:19] Makyo: hmm, I had it in test server. Maybe my system is possessed today [16:19] Makyo: latest trunk? [16:20] rick_h_, yeah, that's where I'm seeing it. [16:20] frankban: yea, no idea. I'm actually newer to the db and not sure of its history. [16:20] Makyo: oh sorry, you did see it [16:20] Makyo: misreading, who put monday in my wed [16:20] rick_h_: no, we don;t have a top level unit list, but the process delta seems to assume we have it. so basically we are calling _process_delta twice on the same model list. trying to confirm [16:20] frankban: heh, well we're half way there then :) [16:20] heh :-) [16:21] * rick_h_ runs for food stuff before next meeting [16:28] redir: sorry, meant to ask that question over here. [16:28] * bac lunches [16:29] redir: over here, over here! [16:30] redir: if it works, please 'bzr add' it to your branch [16:30] redir, ok if not this branch, then share it somehow [16:32] bac: I am in a maze of twisty little passages, all alike.... [16:32] I didn't complete it. I got as far as importing a couple things and looking for the necessary bits. [16:33] but decided that it wasn't trivial and was sorting socks instead of working on the actual card [16:36] bzr makes me feel like a ten left thumbed person [16:51] anyone available for a review? [16:51] hatch: I can in about 5 if you need [16:51] ok just writing docs right now [16:58] jujugui https://github.com/juju/juju-gui/pull/277 whoever is available for a review/qa [17:00] hatch: hit a conflict against trunk, please make sure to rebase off develop [17:00] doh [17:00] small simple one [17:02] rick_h_ fixed [17:02] github thinks I did these commits 6 days ago heh [17:02] I must have broken something in the rebase [17:07] doh lint [17:07] and I just ran it too [17:20] so when I booted up my computer today, refind wasn't there, and neither was grub [17:20] :/ [17:20] hatch: you did a osx upgrade? [17:20] hatch: every time it upgrades I have to reinstall refind [17:20] and then it's all back [17:20] I did some updates [17:21] yea [17:21] ohh [17:21] sonofa [17:21] rerun the refind install.sh and you'll be fine [17:21] ahh cool cool [17:21] osx is a tricky bugger [17:21] it doesn't like the parasite os on there [17:23] refind? [17:24] redir: let's you setup a boot manager to run ubuntu on apple hardware [17:24] http://www.rodsbooks.com/refind/ [17:24] hatch: comments done, a few questions. Let me know if you want to chat on them. Will start QA [17:25] cool looking [17:25] * rick_h_ runs away before he can read them [17:25] does qa from the camper in the driveway [17:25] haha, it's still too cold here for normal camping [17:25] does your camper have a heater? [17:25] I think we've had our last frost warning to trying to get out for mother's day weekend [17:26] oh yea, propane furnace [17:26] and electric heating pads for the bunk ends [17:26] since the tent ends get a bit chilly at night [17:31] lol qa'ing this with the simulator turned on is nuts :P [17:31] yeah it's super broken hah [17:31] that's why I bolded that comment :) [17:32] Uncaught TypeError: Cannot read property 'add_unit' of undefined [17:33] I think my comments in there that hit on this are valid though. [17:34] hmm when did you get that? [17:34] hatch: after the simulator reset the thing 5 or so times and I finally got the numbers input and hit add units [17:34] oh....well yeah, you need to turn the simulator off, the machine view panel is just blowing it all away [17:34] I'm surprised more doesn't break [17:35] :) [17:35] well, let's shoot for less breakage even with the simulator :P [17:35] well it's not the simulator, it's the rerendering of the panel [17:36] if you hack the machine panel view code to stop re-rendering then you don't run into any issues [17:36] I thought that was in here? [17:36] no that's what huw is working on [17:36] the machine panel view listens to any change in the machine model, then re-renders everything [17:39] hatch: right but why does the scale up widget need to listen to those? [17:39] hatch: I think having those move from under the user would be a bit not-good [17:39] it doesn't [17:40] it's in the machine view [17:40] so when the machine view gets re-rendered it re-renders the scale up view [17:40] app/widgets/service-scale-up-view.js - line #70 in the diff [17:40] yeah that's where it listens to re-render the service list if it gets updated [17:41] right, but that seems bad UX imo, a user is there typing and stuff moves on them [17:41] it doesn't if the machine view doesn't re-render [17:41] that's what the _updateUI method does [17:41] and why it's so funky [17:41] ok, let's forget the machine view panel and just look at the life cycle of this new View [17:42] it's init'd and rendered, part of that is binding to change events on the db.services.on() [17:42] and if that changes, adds, removes, it updates the service list, which updates the UI [17:42] all while the user is trying to enter a 3 into the 4th box... [17:43] which is now the 5th box, or doesn't exist any more because of the change event from within the scale up view itself [17:43] well if the service goes away then they can't scale it up anyways [17:43] if the service is still there, it doesn't modify the UI [17:44] so you'll still be focused and typing in the input [17:44] that's horrible to users, and order it promised to never move the list on you? [17:44] if I'm entering in the 4th box and the 2nd one goes away? [17:45] I don't think we can treat this widget as live updating. [17:45] You thought the other way in Vegas :) [17:45] heh, well I think when you click on the + is needs to be live [17:45] and when you close it/reopen it it's updated [17:46] but only at interaction points [17:46] so what if they scale up a service which doesn't exist anymore? [17:46] then we'll have to be smart and bring up a service [17:47] or throw an error during the deploy confirmation step [17:47] so now adding a unit deploys a charm? [17:47] we've got places to confront the user without moving their UI around on them as they try to type [17:47] :) [17:47] we'll probably have to start with the confirmation error, we need that anyway in case of conflicting names/etc [17:47] but yea, I should be able to say "oh I wanted a unit of that, sorry rick blew it away, I want it back" [17:48] "rick doesn't need mysql any more but I sure do" [17:48] then you need to go deploy it again the usual way [17:48] having a scale up UI deploy charms is a little crazy [17:48] ok, still. Regardless of the 'how' I think that once you put the form in front of the user you can't go moving it [17:49] if you're entering a username someone just took, I don't delete your data entry, or delete the field. [17:49] I show you an error/message to help you resolve the issue [17:49] and if a new service is deployed you want to mass scale up you have to click the X then the + again to see it? [17:49] and then this code gets a lot simpler. It doesn't need to watch the db and such [17:50] if you click + and someone adds a service you weren't planning on scaling it already anyway [17:50] Iunno, I disagree [17:50] it wasn't there when you started, you might want to go back and scale that one up before you hit deploy on the bar, but you weren't planning on touching it at the moment you started your interaction [17:51] how could you know it was there? [17:51] going to be there I guess [17:51] it's just odd that it'll be the only place in the UI that doesn't maintain env state [17:51] the inspector unit list jumps around while you may be working on ti [17:51] does the inspector config auto reload if someone does a service upgrade? [17:52] right, and everyone hates that darn jumping unit list thing [17:52] it's so hostile to users [17:52] well, it's accurate [17:52] and this is a small interaction with very very limited env interaction. It's only giving you an option to do something based on a list. [17:53] hatch: sure it's accurate, doesn't make it nice to use [17:53] I'll remove the binding stuff - but maybe the X should turn to a refresh button or something if the db changes [17:53] hatch: hey, I'm all for getting other opinions. [17:53] it's not very intuitive if the user is living in the machine view and wants to mass scale up and there are no services in the list anymore [17:53] hatch: I'm just sharing mine, I think the code can be simpler and easier to use without the event watching [17:54] oh for sure [17:54] if the users live in machine view I think we've failed imo [17:54] but that's a different discussion [17:55] which we can have in our upcoming 1-1 :P [17:55] hah, well I'm totally confused of the direction of the new no-inspector UI [17:55] which may mean people live in there [17:55] you and me both. Had a chat with luca bout that yesterday [17:56] jujugui quick review/qa: https://github.com/juju/juju-gui/pull/278 [17:56] I'm hoping he changed his mind? :) [17:58] Makyo I think that your branch needs to redirect the user to the default of sectionA [17:59] thoughts? [17:59] so their url actually changes I mean [17:59] +1 redir like we do for /fullscreen and such [17:59] Okay. [17:59] to end up in a good place [18:00] wha? [18:00] Makyo that may mean that it'll need to be done in the 'validator' which is yet to be written [18:00] redir I think rick_h_ meant to say me or Makyo :) [18:00] redir: huh? [18:01] hatch, +1 on redirecting, but if the validator has yet to be written, that's outside the scope of this branch. [18:01] oh redir as in short for redirct [18:01] haha [18:01] :) [18:01] sorry, poor naming choice there redir :P [18:01] long running joke [18:01] Makyo yeah, definitely, so would that mean this branch needs to be benched? [18:01] Or land as-is and fix later? [18:01] I guess, now I'm wondering...give redir a task it ends up on Makyo. like a fine waxed car [18:02] lol [18:02] hatch, Either or. I'm loathe to have wasted time on it while things diverge further. Can we add an XXX comment and land? [18:05] yeah that should be fine [18:17] if robbie *just discovered* juju.academy does that mean he slept through the lightning talks? [18:24] bac haha [18:27] Oh, neat, we actually got that URL. [18:29] Have it up on the book site now: http://book.exploring-juju.com/ [18:30] nice :) [18:30] look at all you guys, being good juju community peoples [18:31] hi guys, is there a way to have a password field in the charm’s properties on the gui? [18:32] alexpilotti do you mean the password for the GUI/ [18:33] hatch: in a new charm, we need to specify among the properties a password [18:33] oh ok, well atm that's just a typical field [18:33] I'm guessing you want it to be a *** while you type? [18:33] hatch: the “service settings” in teh GUI [18:34] hatch: yep, typical password edit field [18:34] yeah, at the moment that's not supported but there have been some mumblings. Would you be able to file a bug requesting this feature? [18:34] https://bugs.launchpad.net/juju-gui [18:34] hatch: sure [18:34] thanks [18:37] bac: test_search.py:TestReindex.test_reindexed_no_client_charms deletes the non text index from the ini. Is that intentional? [18:38] redir: don't know. let me look. [18:38] hatch: done: https://bugs.launchpad.net/juju-gui/+bug/1317228 [18:38] <_mup_> Bug #1317228: Passwords fields should be supported in the charm service settings [18:39] hatch: is there a way to hack this in a quick and dirty way? We just need it for a demo [18:39] hmm [18:40] lemme take a look [18:44] redir: i don't understand that test. [18:44] bac: me either [18:46] alexpilotti: hatch this should be on juju. They need to support the idea in their config.yaml [18:46] redir: but this is why es data all disappear... [18:46] I guess it can be on both [18:46] bac Yes, AFAICT. [18:46] rick_h_: I agree [18:46] alexpilotti: what's the demo? Can you tweak the list of config values so the password is at the end/off the screen? [18:47] rick_h_ I've also shown him where we render the config UI and the template in PM [18:47] alexpilotti: or hard code the value in the fork of the charm and remove it from the config [18:47] rick_h_: we’ll do that if we have no other chance :-) [18:47] redir: if you have the cycles, would be nice to ingest some stuff, disable that test, run make test, and then ensure your index is still there. if so, perhaps just leave that test disabled until we can re-write it sanely. [18:48] alexpilotti: ok, we're up against a wall for ODS demo atm so not a lot of time until after next week to look at better solutions [18:48] bac affirmative [18:48] those are the things I can think of off the top of my head to get by for the moment [18:49] redir: and if it solves the problem then i'll buy you a beer at our next meeting. [18:49] ok stepping out for lunch [18:49] bbl [18:59] bac was there a bug for that or just a general annoyance? [19:00] latter [19:11] merge problems:/ [19:11] bac: I should be doing something like "bzr merge ../trunk" yes? [19:12] redir: do you have a current trunk branch? if not: bzr merge lp:charmworld [19:12] I went to trunk and did bzr pull [19:12] but I'll try merge lp:charmworld [19:13] and it is on revno 508? [19:13] redir: what was the merge problem you saw? [19:13] yup [19:14] some conflicts that seemed easy enough, but I wind up with a mongo_services object that is undefined [19:15] redir: sounds like bad conflict resolution [19:15] compare api/__init__.py with trunk [19:19] rick_h_: where are the docs on how to set up tmux like you have it, where it just auto starts with tabs? [19:20] bac yeah that doesnt' seem like it merged very well at all [19:21] and test failures. lemme try again [19:27] jujugui: http://qa.manage.jujucharms.com/heartbeat is alive again, after a long hiatus [19:27] bac: \o/ [19:28] rick_h_: let me know if you have a moment to talk about my branch… at a decision point here. [19:28] turns out, some services on canonistack using orangesquad credentials did not specify admin-secret or control-block in environments.yaml. juju made up strings for those values, but used unicode values that swift could not handle. thus swift went crazy for everything associated with orangesquad. at least that's how i understand it from sinzui's telling. [19:29] bac: help me out here… what's that do and what changes now that it's back up? [19:30]