[00:16] jujugui lf a review/qa https://github.com/juju/juju-gui/pull/204 plz and thanks [00:16] ^ huwshimi if you have a chance today because this will then allow you to style the inspector to fit in the sidebar [00:17] hatch: I'll take a look [00:17] thanks [00:18] If it's all good can you :shipit: for me plz? [00:19] hatch: looking [00:19] DOH ci failure [00:19] doh! [00:19] and I JUST ran lint too lol [00:19] so much fail [00:20] almost done [00:20] NFS makes lint and test running so much faster [00:20] run the full test suite in ~20s now [00:20] would be even faster if it wasn't for the 404's [00:21] fixed [00:24] hatch: What's the easiest way to get your branch? [00:24] For QA [00:24] huwshimi at the bottom there is a link which says 'command line' [00:24] check out develop and update it to the most recent version [00:25] hatch: At the bottom of what? [00:25] then run 'step 1' from the thing which opens [00:25] the PR [00:26] huwshimi https://www.evernote.com/shard/s219/sh/86a14db3-0e37-4522-8be6-5958ad3e2a1a/63d67cfd49665ea37786a882fc3b78f9 [00:26] but just do step 1 [00:26] and make sure you're on develop when you run the commands in step 1 :) [00:33] hatch: I don't know enough to do a code review, but unless something is broken on my end it is not qaing very wel... [00:33] *well [00:33] no? [00:35] what is it doing? The inspector won't render properly in the sidebar if that's what you mean [00:35] hatch: So it's ok for it to appear above the browser content etc? [00:36] there should be no browser content [00:36] hatch: Well, I can get into a state where both are visible... [00:36] hatch: Want a hangout? [00:36] yes plz [00:37] wow that is fugly lol [00:37] haha [00:38] it's just not centered :) [00:39] ugh, so it renders under the editorial [00:39] so remove the top/right nad it starts to fit [00:39] hatch: can you override those in your branch at least? [00:40] .yui3-juju-inspector left/top hard coded values [00:40] and position abs [00:40] yeah I can - I just left that styling alone [00:40] hatch: yea, it's just a small set so it's more qa'able and such [00:40] set them to inherit and it 'works' [00:40] hatch: https://plus.google.com/hangouts/_/76cpi08kvdqlf8ok5huv2frop4?hl=en [00:40] I just want to see huwshimi's issue because it should destroy the charm lists before rendering the inspector [00:41] hatch: yea, it didn't [00:41] hmm [00:41] hatch: I'm betting because it doesn't have a search result it defaults to a editorial list [00:41] riight but locally it does [00:41] hmm, the list is empty, but it's there [00:41] bws-editorial [00:41] has three empty divs in it in my test [00:42] odd it calls [00:42] if (editorial) { editorial.destroy(); } [00:42] if (search) { search.destroy(); } [00:42] before rendering the inspector [00:42] so something else must be broken there [00:47] rick_h_ ok the issue huwshimi found is because we aren't relying on the state to render things into the sidebar [00:47] but that still doesn't explain why the destroy calls aren't working [00:47] hatch: well they're doing some work because the containers are empty [00:48] hatch: but something is still providing default empty containers [00:48] hatch: I think something is redrawing it, again because the state says so [00:48] entirely possible I suppose [00:48] hatch: you're probably still hitting the sidebar() function because nothing is saying not to [00:49] right - we need to sit down and outline every possible state combination [00:49] and map the flow of the state rendering [00:49] hatch: right, that's what kadams and I are doing [00:49] oh cool [00:50] hatch: we're picking up the state branch and started to pull that out today. We should have an initial 'refactor of current state' tomorrow and can start to update it for the new stuff we need after that [00:50] ok great - so after all the xxx's and the little css cleanup this can land? [00:50] it's really broken without the state :) [00:50] hatch: yes, ok from me [00:50] hatch: yes, and that's the reason it can land lol [00:51] haha [00:51] hatch: but yea, it's the highest priority atm and we'll get it unblocked and going [00:51] but with that we can ask huwshimi to start to style the inspect in the space :) [00:51] inspector that is [00:51] sure thing - so next card for me in the AM? [00:51] though a working tabbed inspector will be good to get in there as well [00:52] hatch: safari CI? [00:52] and removing the safari blocker [00:52] hatch: ideally we'll do some releases tomorrow, not sure if we'll get it then or monday atm [00:53] hatch: and it'll give us tomorrow to try to get some state work landed [00:53] ok sounds good [00:54] hatch: let me know if you need to chat on the CI thing in the morning. [00:55] sure thing - I'll probably take a look again to remind myself what the issue was [00:55] hatch: yea, I never figure out what the issue was either [00:59] huwshimi there is a card called 'update new inspectors UI to fit within space of the sidebar...' You can probably use that for your styling [00:59] or part of it [00:59] it's a 4 so 2 days worth [00:59] hatch: OK. [01:00] rick_h_ I'll remove yours and kadams head off of the safari ci card [01:00] and put mine there [01:00] hatch: huwshimi but that's for both of them and some updated UI from luca [01:00] hatch: yep thanks, we were going to work on that today but the state stuff is higher priority [01:01] after the CI is up does safari get let in? [01:01] I think that was our last safari issue [01:01] hatch: yep [01:01] that's the last one [01:01] got it === marcoceppi is now known as marco_traveling [12:54] benji: morning, I'm going to try to do a deploy of charmworld today. Anything on the bundle search issue that I should hold up on? [12:54] rick_h_: given that tomorrow is Friday, I wouldn't hold the release [12:54] benji: ok, thanks [13:08] * frankban lunches [13:32] morning all [13:39] morning hatch [13:49] hatch: that's a cute link there [14:01] rick_h_ :) [14:04] -13C out with a high enough UV index to get a tan right now [14:04] lol [14:05] we're around 0C with rain that can't decide if it's wet or frozen [14:08] worst! [14:08] that's like what it's like on the coast sometimes - that's why I don't live there [14:08] hah === benji_ is now known as benji [14:09] if it's raining it's got to be above at least 13C :) [14:09] tan or windburn? no one can tell [14:09] see what I did there with the 13's [14:09] haha [14:09] nope not windy at all because the temperature has been pretty stable [14:10] the best kiting wind is in spring and fall though when there are big swings in temerature [14:10] temperature event [14:10] even [14:10] ugh [14:18] sublime text users, I found a cool window manager https://sublime.wbond.net/packages/Origami [14:19] a window manager for your editor? [14:19] * rick_h_ shakes head [14:19] tiling manager? [14:19] pane manager [14:19] lets go with pane manager :) [14:19] lol [14:33] jujugui manage.jujucharms.com is down [14:34] hatch: yep, working in #webops on it now [14:34] bad merge on upgrade [14:34] ohh [14:34] we need a staging server :P [14:35] well the cowboy was on prod and not staging so not sure how it'd help atm [14:35] ohh right I forgot about that [14:36] warned them the cowboy was there but it still went boom :) [14:37] safari failure could be an ssl issue [14:37] hatch: k, interesting and good to know [14:37] hatch: hmm, what ssl are we touching? I didn't think it was ssl'd on there [14:38] there are boneafied failing tests though too [14:38] when I run the test through sauce labs there is a safari alert about the ssl not being correct [14:38] hatch: duped locally? [14:38] hmm, ok wonder what ssl thing it's hitting. [14:41] might just be an aws thing [14:41] https://www.evernote.com/shard/s219/sh/72f2a602-ee2a-4ecc-9385-54aa3bc0941f/337790dc870b72c0bd1cdc6480bece24 [14:42] hatch: hmm, can you look for an avenue to hit up saucelabs about this? [14:42] hatch: irc or bug or something? [14:42] as you're right, that's before it loads up on our url we're providing them [14:43] at least I don't think that's our websocket [14:44] we run a lot of tests [14:44] heh, trying to find the old ones [14:44] hatch: heh yea [14:46] interesting...the failures were intermittent before failing all the time [14:47] yup looks like it was the ssl issue https://saucelabs.com/tests/7fc11958a9ff4f859b5b17e79a777f39 just hangs on the 'SafariDriver Client' page but I'm assuming the alert is above the video layer [14:50] jujugui call in 10 please kanban [14:53] hatch: IIRC we use secure=false when deploying the GUI for tests, so, no SSL handling should be required. Looking at the jenkins failure, it seems there is a problem instantiating the remote webdriver (http://%s@ondemand.saucelabs.com:80/wd/hub), but not sure [14:55] benji: do you recall what the production ingest is these days? Is it still the supervisord run process? [14:55] frankban trying http [14:55] frankban same issue [14:55] rick_h_: I think so; it runs forever (and has enqueueing integrated into it so there is just the one job) [14:56] hatch: yeah, the problem seems to be in setupClass, so before even trying to fetch a browser URL [14:56] benji: k thanks for the sanity check [14:58] its redirecting to https [14:58] even when being given an http url [15:00] oh I had secure=true [15:00] heh [15:00] I wonder if I'm in the right hangout. [15:00] jujugui on a call and running late [15:00] hatch: can you run please? [15:01] and I'll catch up in a few min [15:21] frankban I've noticed that when I use quickstart on amazon it takes a v.long time to bootstrap - thoughts on putting a message "waiting for provider" pinging the console every 30s to let the user know it's not hung? [15:21] or something to that effect [15:22] hatch: what's the last quickstart message before hanging? [15:22] I closed it....but it was something to the effect of "...Bootstrapping..." [15:25] feel free to tell me I'm stupid :) [15:28] hatch: they introduced synchronous bootstrapping in 1.17.x. This is why it takes some time. In previous versions, the "retrieving environment status" was the slower bit. I am sure the goal to notify something to the user is worth the effort of spawning a separate thread to ping the terminal. One idea could be printing the synchronous "juju bootstrap" output to the console, but it contains some tecnical data not rea [15:28] lly interesting for the quickstart user I guess [15:29] Yeah - My only concern is for the user experience, so that they don't think it has hung as sometimes provisioning on the providers can take a while [15:29] hatch: what about a slack card to investigate those ideas? [15:30] yeah I'd be fine with that [15:30] slack card in the slack pit [15:30] lol [15:30] :-) cool [15:34] done [15:38] guihelp Just submitted PR https://github.com/juju/juju-gui/pull/207 for refactoring state out of browser. As noted in the PR, it's just the first round; rick_h_ has plans… evil plans… for subsequent cleanup. [15:38] lol [15:38] on it [15:38] Thanks [15:42] rick_h_ https://saucelabs.com/account it looks like it's running the selenium tests twice for each browser? [15:59] kadams54 review done [16:02] rick_h_ I think we should prioritize removing those http requests from the tests - the test runs now take over 40mins with safari in there :O [16:03] maybe timebox it at 1 point? [16:18] rick_h_: James page raised a concern re: quickstart in main/juju-core not in main. Please take a look at #juju when you are available [16:22] frankban: looking [16:22] rick_h_: thanks [16:27] hatch: yeah, I'd like to see save() and update() disappear from the API entirely, but even if that isn't possible, maybe we can at least simplify. [16:27] kadams54 yeah - this state stuff is super confusing for pretty much everyone but rick_h_ :P So any way we can simplify it the better hehe [16:27] * hatch figures that's how rick_h_ felt about the viewlet stuff [16:27] :) [16:27] :) [16:27] lol [16:28] like I said in the interview, what you throw comes back at you. We all have different brains [16:28] OK… so you all saw our seating arrangement in standup… rick_h_ is behind me. [16:28] hahaha [16:28] Considering his status as a ninja team lead, I find this very disconcerting. [16:28] I need to get in before him and grab the back desk. [16:29] Have my back against a wall and all that. [16:29] that still leaves you exposed from above [16:29] maybe also be UNDER the back desk [16:30] If I see rick_h_ scuttling across a ceiling, exorcism-style… I'm outta here ;-) [16:31] rick_h_ I think CI is borked [16:31] might need to restart the vm [16:31] http://ci.jujugui.org:8080/job/juju-gui/630/console [16:31] Is that why my merge build failed? I was having a WTF moment… [16:32] yeah EADDRINUSE [16:32] hatch: sec ,just have to kill the process [16:32] ok cool [16:32] give me a minute, getting hammered from 4 sides [16:33] I can try it [16:33] I've never done any the azure stuff [16:34] ohh I need the env details [16:34] ok I'll just wait :) [16:38] process killed, try again hatch kadams54_ [16:38] trying again [16:38] rick_h_ did you get kadams54_ creds for Jenkins? [16:38] hatch: hmmm, probably not [16:38] :) [16:39] I can queue his up for now [16:39] lol [16:39] or if you wanted to show him [16:39] hatch: thanks, yea busy atm appreciate the help [16:39] I have access to Jenkins [16:39] oh ok cool [16:40] go to the juju-gui project [16:40] go to 'Build with Parameters' [16:40] Hmm, I don't see that… [16:40] and type origin/pr//merge [16:40] ok then you aren't logged in [16:40] right, he needs a login [16:41] Ah, yes. [16:41] ok I'll queue yours up [16:41] you can learn later :) [16:41] Thanks [16:42] queued [16:43] now I need to figure out how to work on this deployer [16:43] python...bzr...sheesh I'm going back in time here :P [16:43] * hatch ducks [16:44] Hah. [16:44] OK, lunching… [16:44] * rick_h_ goes for lunch afk [16:47] this is a pretty cool technique for styling selects http://www.bennadel.com/blog/2594-Using-Transparent-Select-Menus-To-Create-Styled-Menu-Roots.htm [17:18] pythonista TDD question: rails/rspec testing has this concept of loading fixture data after mocking out calls to an external api. I assume to achieve this i just import json and create a var with json.loads(open file descriptor) to get this? or is there a module I should be using [17:18] because python testing fixtures appear to reference a different area of the harness [17:25] frankban still around? [17:25] hatch: yes [17:25] got a second to give me a hand getting the juju deployer running in a virtualenv? [17:26] hatch: sure [17:26] awesome, so I followed the Installation instructions in the README exactly [17:26] (there are only 3 steps) :) [17:26] running it on Raring in vagrant [17:27] and now when I try and run it I get this error https://gist.github.com/hatched/b5f95ae4b378307f8690 [17:27] sorry, Saucy I mean [17:28] it says it's missing a module so I'm guessing the installation is missing a step [17:29] hatch: yes, maybe, apt-get install python-bzrlib [17:29] installing [17:30] u da man [17:30] thx [17:30] :) [17:30] I'll add that to the docs [17:30] hatch: yes, it should be added, thanks [17:31] oh awesome, the tests don't pass [17:31] rofl [17:31] oh... error? [17:33] ahh it needs bzr to run the tests [17:34] I really like how easy python is to read [17:34] the tracebacks are pretty unhelpful, lots of indirection [17:34] but that just might be how it's written [17:36] oh boy the tests actually pull from a remote repo [17:37] 1 failure 2 errors [17:37] getting better [17:41] 1 failure [17:41] blarg [17:42] hatch: bzrlib is a crazy dependency and will give you random failures on clean environment building for testing. [17:42] it proxies through 6 locations to fetch the package [17:42] lol [17:42] the random failures are thanks to one of those proxies timing out [17:43] so I should blame kapil for using a test which goes over the wire? :) [17:43] annnd Safari is back running on CI [17:43] boo yeah [17:43] hatch: this is exactly what I was talking about earlier. I need to mock an api request [17:43] and load "fixture data" to mock the response. [17:44] eg: stuff json in the mocked response so the code i'm testing continues with fake data. [17:44] yeah I can't help you there - with js, no problem - with Python, no idars [17:44] * lazyPower dies a little inside [17:44] i had so much hope hatch, you got my hopes up and dashed them [17:44] lazyPower: the mock library works well for those needs [17:45] * rick_h_ is back from lunch [17:45] frankban: im' mocking the request.post, can you point me for how to do the response? should i just json.loads(data) and assign it to response_object.data.return_value? [17:45] is that the proper method? or is there a module/library for helping do all of that setup? [17:46] lazyPower: to be clear, I am referring to http://www.voidspace.org.uk/python/mock/ [17:47] frankban: from mock import patch, Mock [17:47] using the same lib, its got magickmock [17:50] lazyPower: in http://bazaar.launchpad.net/~juju-gui/juju-quickstart/trunk/view/head:/quickstart/tests/test_utils.py#L681 there is an example of a way to mock urllib.urlread [17:50] rick_h_ so I'd like to prioritize those http requests in the tests if we could...the CI is getting long in the tooth and causing timeeout issues now that Safari is in there [17:51] frankban: you're my new hero, thank you! [17:51] lazyPower what is your language of choice? [17:51] hatch: create a bug and card in maint. Where's it causing time issues? I thought we had 10min windows on each sauce run. We're no where near a 10min run [17:52] rick_h_ sometimes the http request just hangs [17:52] no idea wy [17:52] hatch: in the on deck backlog please [17:52] sounds good [17:52] hatch: right, but do we have evidence that mocking the http requests avoid this? [17:52] lazyPower: welcome. Hope you find it helpful to obtain the same for whatever call you need to mock [17:52] I see dots, do we have anything to point to that connects them? [17:52] rick_h_ well if it doesn't make the http request it can't hang on it :) [17:53] hatch: ok, so where is the failure hanging on the http request? [17:53] lazyPower: in general, quickstart interacts with external data sources a lot, and so mock is used pervasively in the tests [17:53] http://ci.jujugui.org:8080/job/juju-gui/628/console [17:54] yuss passing juju-deployer tests [17:54] ok now I can get to work [17:54] well it's lunch [17:56] this of course was just the most recent failure [17:56] hatch: yea, I'm just trying to find evidence of the connection [17:56] hatch: are you saying this is the safari issue? Or is that just a side note here? [17:57] oh no it's just an issue - but because we now also have safari test runs are nearing 40mins [17:57] that's getting a little too long [17:57] imho [17:57] hatch: right, ok but that's not causing failures. That's different [17:58] well that one I linked you hung on that http request [17:58] ok that's fair I guess. last I looked we were at 26min with safari, will see [17:58] hatch: righ but it's the last line of output, what indicates 'hung'? [17:58] hatch: what makes this obvious vs a sauce issue, or something else? [17:58] I had to kill it at like 50mins which means it was probably sitting on that line for 20mins :) [17:58] right [17:58] our end does not handle sauce going away === marco_traveling is now known as marcoceppi [17:59] so that time isn't a real time [17:59] so your merge that landed: Took 27 min [17:59] per jenkins [17:59] right I suppose [18:00] :) just keeping you honest [18:00] lol that's fine [18:00] ok, I'm for fixing things but it feels like we've got 1) safari issue, not an issue now? 2) Tests take longer with 4 full browsers, we should fix 3) intermittent failures, cause unknown [18:00] safari issue appears to not be an issue [18:00] or is there more direct info and ^ is wrong [18:01] it 'just works' [18:01] lovely [18:01] so Safari is now unblocked as well [18:01] no more warning [18:01] hatch: oh, did you land that as well in your diff? [18:01] yeah it was so trivial of a change [18:01] hatch: awesome ty much! [18:02] I added a test for it too so it's all good [18:02] np [18:02] we should focus some QA'ing on it though [18:02] woot [18:02] just to be safe [18:02] I can help with that [18:02] hatch: yes, it's a friday card [18:02] oh cool [18:02] hatch: and I'm not going to do a gui release this week [18:02] hatch: I'm going to wait for now, we only released charmworld/quickstart this cycle [18:02] probably a good idea - it's a little in flux atm :) [18:02] yep, agree with you [18:03] * rick_h_ takes meds for agreeing with hatch :P [18:03] lol!! [18:03] hatch: did you get a chance to re-queue my PR merge build? (#206) [18:03] kadams54 yep I think it's running now [18:03] sec [18:05] kadams54 commented in the PR [18:05] you can shipit now [18:05] kadams54 oh it's running again because trunk changed [18:06] http://ci.jujugui.org:8080/job/juju-gui/633/ [18:06] oh well, I'm sure it'll all be good :) [18:06] lol [18:07] all good, ship ship ship! :) [18:07] Yeah, rick_h_ just pointed out that build #632 was for PR 207; just didn't show in the Github conversation since it was kicked off manually [18:07] yeah we should probably fix that [18:07] Since I've seen at least one good CI build, I'm fine doing :shipit: [18:07] ....todo [18:07] :) [18:07] Assign it to the new guy [18:07] Oh… [18:07] Wait… [18:07] lol [18:08] I definitely want work done on the test suite - just need to convince people to stop assigning us feature tasks :P [18:09] I've been telling rick_h_ the same thing! [18:09] We tried to convince Mark R. yesterday. [18:09] He was less enthusiastic ;-) [18:09] haha [18:10] hatch: hey, you knew machine view in a hurry was coming [18:10] it's #1 because honestly we're less than a month from vegas and they'll want to see it almost ready there [18:10] hatch: so I'm +1 on the idea, but let's remember where we're at :) [18:11] while() { printf("you knew %s in a hurry was coming", $project); } [18:11] hatch: now that's not fair. I recall weeks of maint at the end of the year [18:11] hahahaha [18:11] :P [18:12] so after lunch I'm going to get going on this deployer stuff [18:12] now that i have it passing the tests [18:13] woot [18:21] rick_h_: I'm stumped; the indexing problem I thought existed actually doesn't and now the unindexed bundle is suddenly indexed on prod [18:22] benji: ummm, ok [18:22] benji: well that's good that the problem doesn't exist. That would be a big use case hole. [18:22] my professional diagnosis is [18:22]