is the renderTo in the code for the target of that event [06:04] but yea, not firing [06:05] yeah, that's it [06:05] rick_h_: .bws-content is the element it should have the event on. [06:06] ah ok, updating [06:06] rick_h_: We could possibly make #bws-editorial the element that scrolls if need be. [06:07] yea, that works [06:07] Oh good. [06:07] I just hard coded it to .bws-editorial for now [06:07] the structure is a bit different, but this will work for now [06:07] thanks! [06:07] rick_h_: No problems. Hope you get some sleep! :) [06:10] I will head off to bed now. Have it pretty close. [06:11] night and thanks again! [06:14] rick_h_: Night [10:49] rick_h_, Morning [10:53] antdillon: morning [10:55] rick_h_, I have a few issues with the environment change set. Where is the best place to voice them? [10:56] rick_h_, Also can I check I should be working from flag mv/li? [10:56] Just me for the moment. Want to chat? Or type out what's up? [10:56] antdillon: yes, this is work under the two flags. It's really mv, but il is the soon to be released 'inspector left' flag [10:57] rick_h_, I got it all working its mainly missing info in the ecs like no time stamp [10:57] antdillon: oh? Ok, well I was just expecting the count to work [10:57] so if you've got more stuff working then why not push up your branch, start a pull request with notes, and we can start to look through the code and help fill in the last bits [10:58] rick_h_, I few events to bundled like deploying a server and subordinate fire the same deployed event [10:58] rick_h_, Wont we need time stamps for the summary panel? [10:58] yes, well containers and machines are the same thing in juju land so that's not surprising. There should be some metadata to help us tell them apart [10:59] antdillon: well, maybe. I'll look at the mocks. I'd consider that more for the changelog, which we're not doing right now since we can't get your previous commits [10:59] Oh I see, I was using the command method to split the events [11:01] antdillon: cool, yea if you start a pull request we can start to go through and help hash over the last details and check it out. [11:03] rick_h_, Sure, just working through the last few changes and there messages then tests :) === perrito6` is now known as perrito666 [11:07] antdillon: yay [12:11] frankban: sorry, completely missed the time [12:11] rick_h_: no worries, ready when you are [12:31] frankban: https://github.com/juju/juju-gui/pull/277 [12:36] * frankban lunches [12:41] this test run isn't going to be pretty [13:05] morning all. [13:07] morning [13:20] rick_h_: got your invite--i have a contractor swinging by the house sometime between now and 11, so i may be interrupted, but i'll be there. [13:21] jcsackett: thanks [13:22] it'll be quick, forgot to shorten from the google calendar 1hr default [13:22] cool. [13:30] rick_h_: daily hangout? [13:32] bac: soryr, had a hangout in the meeting invite [13:33] https://plus.google.com/hangouts/_/canonical.com/charmworld?authuser=1&hceid=cmljay5oYXJkaW5nQGNhbm9uaWNhbC5jb20.7enijh4dvef97a4ifdu83bnoj4 [13:50] I <3 it when a passing test comes together [13:52] redir: when you finish your current work, could you look at the deploying local charmworld instructions and walk through them to ensure they still work? you have fresh eyes. https://docs.google.com/a/canonical.com/document/d/1Y8Uhomr4_6L3nFXLXPZY2V-tFjl5KjOIknQLuXdzU_A [13:53] bac do we have a target for the demo? trusty? [13:53] I can do it from scratch in a container [13:53] yes, i'd say trusty [13:54] in an lxc would be great [13:55] k [14:08] bac: want to catch up with 1-1 later on? [14:09] rick_h_: what do you mean? postpone? i can do it now [14:09] bac: ok [14:21] Hi guys, does anyone have a min to help me get my head around some tests for deployer bar? [14:24] Makyo: are you around to help antdillon? [14:24] antdillon: I've got a call in 7, what's up? Maybe push a branch and create a pull request with notes/questions so we can see what's up? [14:24] and I can help out between calls [14:25] rick_h_, Its mainly just trying to add a change to the ecs in the test [14:25] antdillon: ah, gotcha. sec, might be able to find an example [14:26] rick_h_, I'm in the text_env_change.js [14:26] antdillon: oh ok, so in the ecs tests itself? [14:27] rick_h_, Something like utils.makeStubMethod(envObj, '_deploy'); [14:30] antdillon: you're looking to make a stub or to put something into that stub'd method? [14:30] rick_h_: fyi, my contractor just called to say he's running late, so he's likely to arrive *during* standup. as with earlier meeting, i'll be there but may get interrupted. [14:31] jcsackett: rgr [14:32] antdillon: maybe hatch can help you 1-1? [14:32] uh oh [14:32] * hatch runs away [14:32] :P [14:33] hatch, come back [14:33] what's up? [14:34] hatch, Getting errors trying to set changes to the ecs in the deploy bar tests [14:34] hatch, Got any tricks? [14:34] can you push up the code? [14:34] so I can see what you're trying [14:34] and what the errors are [14:34] hatch, Here is the test: https://pastebin.canonical.com/109870/ [14:35] Getting TypeError: 'undefined' is not an object (evaluating 'ecs.changeSet') [14:36] looking [14:36] hatch, Not sure im initialising the ecs properly [14:36] blah 2fa, one sec [14:38] ok you're doing the stubbing incorrectly [14:38] one sec [14:38] at least I think so [14:38] you're stubbing out the entire application [14:39] probably, seems slightly different test by test [14:39] so what is this test trying to do? [14:40] when you simulate a click on the deploy button the ecs commit method is called? [14:40] The "should increase changes when a services is added" is trying to set a change then check the deployer bar is returns 1 changes [14:40] oh that test [14:41] woops :) [14:41] I didnt write that one, im working on the one above and will move to that one [14:41] hatch, But both are erroring [14:43] ok so first of all you're stubbing out a method and not resetting it [14:44] and then as far as the rest of the errors I'll need to run the code to see the actual error [14:44] hatch, I'll push them up [14:45] thx [14:45] brb [14:48] merge proposal submitted in lp. requested R=jcsackett to get different eyeballs on it. [14:49] bac: ubuntu or ubuntu-cloud for the container? [14:49] matter? [14:49] redir: lxc? [14:51] hatch, https://github.com/anthonydillon/juju-gui [14:53] antdillon which branch? [14:53] oh you pushed into develop [14:53] Pushed to my fork [14:53] git@github.com:anthonydillon/juju-gui.git [14:53] redir: ack, i'll take a look after standup. [14:54] redir, bac: we're using LP, not rietveld now? [14:54] antdillon yeah ok, give me a few minutes to pull this down and take a look [14:54] jcsackett: new hires cannot use lbox [14:55] something to do with the auth dance [14:55] and unsupported tools [14:55] bac: lovely. [14:55] ok, no worries. [14:55] hatch, Thank you very much [14:57] jujugui call in 4 [14:57] hey rick_h_ you're sending out gcal invites in pacific time. perhaps you should tell google you're back home. :) [14:57] not that it matters... [14:58] bac: hah, awesome [14:58] bac: hmm, it says I'm in eastern [14:59] our 1:1 came across as 7am, which caused me to panic a bit [14:59] antdillon assert.equal(view._getChangeCount(ecs), 1); [14:59] you didn't pass ecs into this method [14:59] hatch: got a state question for you, maybe after standup? [14:59] sure [15:00] jujugui call now [15:03] hatch, Isnt ecs a global? [15:04] hatch, Ah got it :) [15:04] jcsackett: bac's instructions for setting up juju-gui to connect to local charmstore with ngrams applies also. [15:06] antdillon :) rookie mistake :P [15:06] hatch, :P [15:09] jcsackett, shared the calendar with you. [15:09] Makyo: thanks. [15:10] redir, bac: do we have to set it up that way? shouldn't the search api endpoint reflect the changes? was planning on just poking at a local charmworld with httpie for qa. [15:11] jcsackett: i think that works just as well [15:11] jcsackett: using juju-gui to see the autocomplete behavior is nice, too [15:11] hatch, Is utils.makeStubMethod(envObj, '_deploy'); the correct way to trigger an event? [15:12] nope [15:12] jcsackett: that is how I rolled. But hooked up the gui to make sure it didn't puke [15:12] redir, bac fair points. [15:12] that will make a stubed version of the _deploy method on the envObj [15:12] looking now. [15:13] you want to fire an event from envObj? [15:13] rick_h_ you said that you have an outline of the demo they want to do? [15:13] Im copying from test_env_change.js [15:14] you're going to have to give me more information :) [15:14] hatch: yes, I've requested access for everyone, waiting on luca [15:14] bac: yes re lxc, there are ubuntu and ubuntu-cloud images. Not sure which they will demo from butvalidating on the same one would be ideal. [15:15] redir: i have no preference. pick one and document it [15:15] k [15:15] and if it breaks, pick the other one [15:15] hehe [15:18] frankban are you taking over huw's branch? [15:19] hatch: no, I was going to see if you would [15:19] hatch: we need to get the service units stuff landed first before anything else really [15:19] yep I will, I just saw his comments, wasn't sure if he was on it [15:19] he was realy helpful to narrow it down <3 [15:20] haha yes [15:20] ok I've got to run to the drug store to buy some drugs [15:20] yay drugs! [15:20] I've gone this long without any, but with all this coughing it's just too hard to do anything [15:20] rick_h_, frankban: so, what branch should i be watching for service unit stuff? i thought just frankbans, but huw's too? [15:20] need some cough juice [15:20] hatch: I am trying to proceed with this branch, so that wasn't a real review [15:21] frankban ok thanks [15:21] bbiab [15:21] hatch: just make sure it's not the *special* cough juice. [15:21] jcsackett maybe the code will be more interesting :P [15:22] yeah, b/c coding with full-strength robo-tussin is such a good idea. or is that like the balmer curve... [15:22] jcsackett: that's my branch: we will have a db.units lazy model list, and new db.addUnits() / db.removeUnits() calls to use in order to keep the different model lists in sync [15:22] hatch, Sorry I was using the code from test/test_env_change.js [15:23] frankban: awesome. you're not doing anything insofar as loading units onto machines, are you? i'm assuming i'll want to have the machine view code drop that onto tokens as appropriate from the lazylist you're making. [15:23] hatch, All I need to do it trigger an event like _deploy or _relation [15:23] s/machines/machine-view/ [15:24] jcsackett: yeah, at that point it should be something like db.units.filter(function(unit) {return unit.machine == $myMachine}) or similar [15:25] frankban: awesome. thanks. [15:40] redir: to make sure i'm understanding, ngram stuff is available in ES, you've set up how you want ngram stuff done with n3_20 stuff in filter and analysis, so ES understands what is meant by that when you use it in the mapping? [15:41] jcsackett: yes. define an analyzer then use it [15:41] redir: just making sure. outside of a full understanding of ES it looks a little magical. :p [15:41] agreed [15:42] I don't claim to have a full understanding of ES:) [15:43] back [15:44] antdillon _deploy or _relation aren't events [15:44] hatch, I mean't adding to the ecs [15:45] ok so you need to add a record to the ecs so that you can test that the count goes up in the deployer bar? [15:45] When I do ecs._createNewRecord('service'); [15:45] I get TypeError: 'undefined' is not an object (evaluating 'record.command.args') [15:46] redir: code is r=me, qa-ing now. can you please file a bug about the test you needed to comment out? [15:46] that means that record.command is undefined [15:46] hatch: another question on the hasChanged stuff… [15:46] antdillon the ecs stuff requires a bunch of setup your best bet is to call one of the public methods [15:46] ecs.deploy for example [15:46] hatch, Tried ecs._createNewRecord('service',{ foo: 'foo' }); with an undefined args.length [15:47] redir: qa may take awhile--i have no charms in my local charmworld, need to let ingest run for a bit. [15:47] kadams54 shoot [15:47] jcsackett: and with that test commented out they won't get deleted when you run the suite [15:47] :) [15:47] jcsackett why does charmwork injest take to long? [15:47] hatch: don't go there :P [15:47] oh [15:47] haha ok [15:48] * hatch steps back slowly [15:50] hatch: right now, the way things are coded up, it doesn't do a deep object comparison. https://github.com/juju/juju-gui/blob/develop/app/views/state.js#L70 [15:51] well that's stupid [15:51] That's a problem for metadata, which, as things stand, will always return true. [15:51] what kind of idiot wrote that [15:51] :) [15:51] I'm pretty sure it's Benji's fault ;-) [15:51] lol! [15:51] So… how do you want to fix? [15:51] I had to do this for filters [15:52] the 'proper' way to do fix is to add dot notation to that method [15:52] but the easy way is ot manually compare for now [15:52] sorry [15:52] * kadams54 *sobs* [15:52] manually compare isn't too bad [15:52] one extra line [15:52] kadams54: hatch I just did JSON stringify compare to check if they were the same for the filter object data [15:52] fyi [15:52] rick_h_: yeah, I was thinking of doing the same [15:52] ahh yeah that would work, but in this case that's overkill [15:52] Couldn't remember if there were any caveats to that hack [15:53] imho [15:53] kadams54: determinism of the data [15:53] it's slow [15:53] "slow" [15:55] antdillon you can probably go ecs.changeSet.abc123 = { ... } [15:56] assuming that you don't have another key which is abc123 :) [15:57] hatch, lol i thought that would be cheating [15:57] hatch, wanted to try and push a true event in there using the right function [15:57] well it is, but you're testing the integration between two classes [15:57] you don't need to test that the ecs is working, it already has tests for that [15:57] i guess [15:57] thanks [15:58] np :) [15:58] I don't know how the stuff in cough syrup works but daaaaaaamn it works fast [15:58] on to the deploy :P [16:00] antdillon so how is the Object.observe stuff working out? [16:01] hatch, Good! seems to be working for the changes that are supported [16:01] nice [16:01] glad to hear it [16:01] should have a pull request in soon [16:07] jcsackett: filed 1317567 [16:07] redir: thanks. [16:18] jujugui need two reviews and would like two qa's of https://github.com/juju/juju-gui/pull/281 please. Apologies for the size up front. took 4 files down to 1 so lots of moving around [16:21] QAing [16:21] kadams54: ty [16:22] hah, and somehting is broken yay [16:23] frankban so in huw's branch the proper add event is being fired but `this.get('db').machines.filterByParent(null);` is not returning the newly added machine [16:23] but there IS a machine in the db [16:23] hatch: I explained what's going on in the mp comments [16:24] ohh woops I misread your comments [16:24] .filterByParent(null) does not return the machine because parentId is not set at the time the event is fired [16:24] hatch: ^^^ [16:24] ok I can fix that, instead of overwriting add, listen to the the 'on' event and modify the machine before it's added [16:24] hatch: my current branch fixes the issue [16:25] bah, make test pass rebase and now it's broken [16:25] hold off kadams54 working on it [16:25] is there a reason why ^ wouldn't work frankban ? [16:25] hatch: interesting [16:26] rick_h_: will do [16:26] it seems like it should [16:26] then we don't have to make custom events for fun :) [16:28] redir: so, qa may take a really long time. ingest blew up b/c of local network problems. [16:28] redir: do you have other things to work on, or is this blocking you? [16:29] nope trying to review local demo instructions and have a bug in the lxc ubuntu image trying to figure out where to file that [16:29] jcsackett: ^^ [16:29] hatch: I am not sure, the add override has been there for a while, but if you think that works I agree it can be better, given we find a way to make it easy for people looking at the models what's going on (i.e. you add a machine, and then you have a different machine). This kind of jumping around can be dangerous when using events. [16:31] ok I'm going to give it a try, see if the tests pass [16:31] jcsackett: I have a copy of the ES index I can put up for you to DL but you may also need the mongo stuff for CW to work right [16:31] idk [16:32] hatch: to be clear, I am worried by the double pain of having 1) an object mutated 2) by a ninja subscriber [16:32] redir: if it's not blocking you, i'm content to just let it finish running properly. [16:32] jcsackett: cool [16:32] redir: not that i think your branch could be at fault, but it's always good to make sure ingest still works. :p [16:32] frankban, yeah it's not very transparent with the ninja subscriber [16:33] is "ninja subscriber" an actual term of art, or just something y'all are going with? [16:33] I just feel that we should be modifying the machine BEFORE it's saved not after [16:33] jcsackett just running with it [16:33] :D [16:33] hatch: dig. [16:33] I see a tweet here somewhere [16:34] I need some program that tweets and G+'s at the same time [16:34] ninja subscribers can destroy your application very fast [16:35] see an attribute would handle this in the setter [16:35] I'll ask in #yui see if anyone has any input [16:36] hatch: anyway, what's so wrong with the custom event? we can also avoid the original one to be fired, so the total event count can be maintained stable [16:36] frankban events aren't discoverable - and custom event's don't follow any convention [16:37] so anyone working with the code will have to somehow 'just know' that they should use this custom event instead of the YUI one [16:38] I suppose we could make the add event silent and then fire an add event ourselves [16:38] hatch: I tried to reuse the YUI one, but found that unfortunately an internal ninja subscriber is already in place in YUI :-/ [16:39] something else is listening to the 'add' event? [16:39] hatch: yes [16:39] and our modifications break it? [16:40] https://yuilibrary.com/yui/docs/api/files/app_js_model-list.js.html#l189 [16:41] that's just publishing that this will fire that event [16:41] but I see what you're saying [16:42] the lazy model list api is linking to the wrong file :/ [16:42] hatch: that;s also adding a default subscriber: this._defAddFn [16:42] yeah [16:43] hatch: but perhaps we can fire add with preventDefault? [16:43] hatch: yeah, that could work [16:43] https://yuilibrary.com/yui/docs/api/files/app_js_lazy-model-list.js.html#l287 [16:43] hatch: so we add with silent: true, we modify the object and then we fire "add" with preventDefault? [16:44] just an idea [16:45] hatch, sorry one last thing [16:45] np, shoot [16:45] kadams54: pushing updated branch now. [16:45] frankban I'll try that approach and see how it works out [16:45] jujugui so https://github.com/juju/juju-gui/pull/281 is updated and looking for eyeballs please. [16:46] * rick_h_ goes out to get some lunch [16:46] hatch, assert.isTrue(stubApp.ecs.commit.calledOnce()); is false but it should defo be being hit [16:47] how are you stubbing it? [16:48] hatch, http://paste.ubuntu.com/7417088/ [16:49] antdillon where you think it should be being called place a debugger right before that line. Then console.log the function and see if you get the stub function or the real commit function [16:50] and don't use isTrue (it doesn't give proper tracebacks) use equal() instead [16:50] hatch, Ok thanks [16:53] hatch: ok so I'll exclude that change from my current work, could you please inform Huw that you are working on that later? [16:54] frankban yeah - I'll let you know if this works, atm I'm having trouble preventing the default soon enough [17:00] frankban looks like this technique won't work without a bunch of hackyness [17:00] so we'll have to use your technique [17:01] hatch: fullAdd? [17:01] yeah [17:02] frankban what if we modified the model BEFORE calling var result = MachineList.superclass.add.apply(this, args); [17:02] is that possible? [17:02] so invert how the new add method executes [17:02] hatch: that can be possible [17:03] hatch: especially because we are using lazy model lists, so we should not require calling instance methods [17:03] hatch, Ah so I had to call stubApp.ecs.commit() to get true for stubApp.ecs.commit.calledOnce() [17:03] frankban that's what I was thinking....ok I'll try that [17:03] that makes sense [17:04] ^ antdillon [17:04] hatch, Is there a way to calledOnce on the none stubbed function? [17:04] no but you can call the original method [17:05] The original method is called by the button click [17:05] https://github.com/juju/juju-gui/blob/develop/test/utils.js#L89 [17:05] well if the original method is called by the button click then clicking the button should trigger the stub to be called [17:06] hatch, Its not for some reason [17:07] antdillon: is the code up? Remember that button clicks are events and async [17:07] hatch, But by debuggering and calling stubApp.ecs.commit() [17:07] antdillon: so you have to find a way to watch for that click [17:07] the calledOnce returns true [17:07] my guess is you're running into an async issue [17:07] so what you need to do is listen to 'after' that click event [17:07] hatch, I thought it might be a race :( [17:07] then test for the once() in there [17:08] so yeah....like rick_h_ said :) [17:08] but with the debugger i ran calledOnce a while after the click and its still false [17:09] antdillon: right, so you created a mock, then you manually called the mock [17:09] what you want to do is create the stub, and watch it, and see who ELSE talks to it [17:09] antdillon: let me know if you want to hangout and I can help look at it. I'm free atm [17:09] rick_h_, Ok that would be good. This is the final bit of tests for now [17:10] antdillon: hangout invite on the way [17:11] antdillon: https://plus.google.com/hangouts/_/grvtyzyt53dbgg2c4lkx6elfwya?authuser=1&hl=en [17:12] frankban inverting the execution order of add() fixed it :) yay [17:13] hatch: very cool [17:13] much happier [17:13] :) [17:13] yeah [17:15] frankban https://gist.github.com/hatched/84a65398d22fc3753ef6 here is the new method [17:16] hatch: cool, could you please change also the unitList one? [17:17] I could, that might be out of scope of this branch though [17:17] this branch already breaks the environment switcher [17:17] lol [17:22] rick_h_ when you have a moment I'd like to pair on this review - there are so many lines moved around it would be nice to have some guidance as to what's new and what's not [17:26] sure thing [17:26] hatch: will ping once done with antdillon [17:26] rick_h_, git@github.com:anthonydillon/juju-gui.git [17:27] yeah no rush [17:32] redir, bac: https://pastebin.canonical.com/109893/ [17:32] post ingest, trying to start up charmworld. [17:33] jcsackett: dude you broke i [17:33] t [17:33] * jcsackett laughs [17:33] sorry, dude. [17:34] i didn't want to. :) [17:34] redir: i'm running tests now to see if we get something informative locally to help you debug. [17:34] `make check`, to be specific. could always clobber something bad in my environment and make your code work. [17:39] Makyo I think I failed with your QA....dragging a service to the canvas on comingsoon doesn't open the inspector anymore....after you deploy one service though, it does [17:39] hatch, boo. Oh well, easy fix. [17:40] Let me do that real quick. [17:40] thanks, sorry about that - I'm not sure how I didn't notice that in QA [17:42] hatch: time to chat? [17:42] looks like kadams54 found I missed the cleanup point whoops [17:42] rick_h_ maybe you should fix kadams54's issues he found in QA first :) [17:43] :P [17:43] welcome back to the dark side [17:43] mohohahaha [17:43] I hate to say this, but huw suffered through multiple rounds of QA when I was working this branch [17:43] He might be good to pull in for a second set of QA ees. [17:43] eyes even [17:44] He had a good knack for breaking this in increasingly complex ways [17:44] it's all good, it's close. I didn't test the inspector integration and it's some missing cleanup on close/change [17:44] rick_h_, Branch up updated git@github.com:anthonydillon/juju-gui.git [17:45] rick_h_, Heading home will be online in an hour [17:45] antdillon: do a pull request please [17:45] antdillon: it's easier then you can see the diff of what changes [17:45] rick_h_, Sure [17:45] thanks [17:47] guihelp: I need two reviews and possibly two QAs for https://github.com/juju/juju-gui/pull/282 (this is the db.units change). I have to go now, but I'll take a look later. Thanks a lot, that's a long diff but should unblock things a bit [17:47] frankban I'll take one [17:47] thanks for the hard work on it [17:47] frankban: i'll take the other. [17:47]