[11:32] regression/stop the line: bug 1157138 [11:32] <_mup_> Bug #1157138: deploying a new service does not update service name in environment view < https://launchpad.net/bugs/1157138 > [11:33] jujugui ^^ [11:34] * benji attempts to duplicate [12:01] benji, gary_poster: are y'all working on the regression? can i help? === BradCrittenden is now known as bac [12:01] I'm looking at it independently at the moment. [12:02] I have a hypothesis that the updateGhostServiceName function which is wired up to a "blur" event (seems odd) isn't being called. [12:03] hypothesis denied! It is being cvalled, it's just not effective for some reason [12:06] bac, hi, has your juju-gui remove_relation branch landed? [12:07] teknico: yes, i believe so [12:07] that was my intent [12:07] bac, I don't see the code in trunk though [12:08] teknico: me neither. submitting again. [12:13] benji thank you for looking at it [12:13] benji did you try reverting to see what revision introduced it? [12:13] gary_poster: yep, trying that now [12:13] cool benji [12:21] teknico: submitting now. not sure what happened yesterday [12:22] bac, great, thanks [12:26] bac, while I'm pestering you, :-) are you also landing the destroy-relation juju-core branch? [12:26] teknico: i am trying desparately. i'm in the review grinder. [12:26] teknico: soon, i hope [12:28] bac, thanks (the interest comes from two branches of mine tailing your ones, and both are likely to conflict) [12:28] teknico: understood [12:30] teknico: my juju-gui branch is in trunk now [12:44] benji, any luck on determining version with the problem? do you want to pair? [12:45] it works in r387, does not work in r431 [12:45] benji, heh [12:45] re. pairing: sure; I hear the coffee machine finishing, let me grab a cup and we can hang out [12:45] ok cool [12:49] 400 is ok... [12:50] 420 is ok [12:50] gary_poster: regular hangout? [12:50] sure [12:52] 415 not 420, sorry [13:31] gary_poster: around? [13:31] ahasenack, yes. btw, I think we are going to produce mockups with your suggested approach and see which one people prefer [13:31] s/prefer/understand better/ [13:32] ok :) [13:32] gary_poster: the same happens when viewing a unit, btw, it say "unit LANDSCAPE" [13:32] ahasenack, yes, as designed [13:32] ahasenack, and that's what we would invert [13:32] gary_poster: but anyway, here is something else. I'm visualizing a unit (screenshot: https://inst-005.virtual-maas.com/charms/charms//json) [13:32] gary_poster: see that link to the charm? local:....? [13:32] ahasenack, cannot access that screenshot :-/ [13:33] gary_poster: oh, sorry, I inverted things [13:33] gary_poster: http://i.imagebanana.com/img/gco9lr5r/Selection_002.png screenshot [13:33] gary_poster: see that link to the charm? local:....? [13:33] ahasenack, yes [13:33] gary_poster: that link points to https://inst-005.virtual-maas.com/charms/charms//json, which doesn't load, looks like something is wrong with it [13:33] maybe because of the local:? [13:34] ahasenack, maybe. That charm page has been lame for some time. Not Landscape specific. Thank you for finding it. Are you willing to file a bug or shall I? [13:34] sure, I can file one [13:34] gary_poster: and I have another question too [13:35] let me prep another screenshot [13:35] ahasenack, fwiw, that's not specific to local charms: http://uistage.jujucharms.com:8080/unit/memcached-0/ [13:35] gary_poster: ok [13:36] gary_poster: so I clicked on "view all notifications" [13:36] gary_poster: partial screen: http://i.imagebanana.com/img/u2t2fp6g/Selection_004.png [13:36] gary_poster: why does it say "Problem with"? It looks ok [13:36] lol ahasenack because we never look at this unless there are problems. another bug! thanks again [13:36] I note there are two places in "juju status" where agent-state is mentioned: one in the list of machines (where "running" is good) [13:37] and another in the list of services (where "started" is good) [13:37] ok, I'll file it :) [13:42] Makyo: our second branch (SetAnnotations) works, please ping me when you are available for an hangout [13:48] exciting discovery: lbox submit and propose send off draft comments you've entered into rietveld, avoiding the race of trying to do send them manually. === teknico__ is now known as teknico [14:05] has anyone taken the stopline bug yet? [14:06] teknico: my destroy-relation bug is now being submitted to juju-core [14:06] hatch: benji and gary_poster are looking at it [14:06] ahh ok sounds good [14:07] bac, great, thanks [14:07] hatch, thanks. we have a fix. getting tests to pass [14:07] if bzr has a bisect it 'should' be easy to find :) [14:07] oh :) [14:07] excellent [14:17] jcsackett: are you here today? [14:29] jujugui call in 1 [14:29] review needed of https://codereview.appspot.com/7885045 <<< benji plus one more [14:29] please :-) [14:31] bcsaller_, starting without you [14:31] gary_poster: will miss today [14:31] ack rick_h_ [14:39] gary_poster: you might want to try this branch - it actually hooks up the StateWatcher. [14:39] https://codereview.appspot.com/7663048 [14:39] wow [14:41] hatch: when your call is done wonder if you can peek at something for me please [14:41] definitely - I also wanted to have a quick chat with you about your paths [14:41] for the charmstore stuff [14:42] hatch: rgr, I actually want to have a bigger overall chat this afternoon once I'm done here at the dealer on things like a global regestry/etc [14:42] ahh - what you doing at the dealer? busted car? [14:43] regular 30 day service for the new thing. but yea...did bust a part/light after running a ditch due to some ice the other day as well. [14:43] nothing bad, but ugh [14:49] https://codereview.appspot.com/7885045 [14:50] I can do it [14:51] bcsaller_: so is there anything I can help with with the deploy test bug? [14:51] frankban, Oops, want to head back to hangout, or see-emily-code? [14:51] hatch: https://saucelabs.com/jobs/cf322461890c4645af4f3f05072a60b8 around 30 sec into the video it shows the issue [14:51] Makyo: emily in 9? [14:52] frankban, Sure. [14:52] ok loading [14:52] hatch: if you have any ideas I'm open, I was trying to figure out if the unit tests time out it could leave things in a state where it then needed the login on the next screen. I might try a run w/o the unit-tests on just to see if its any different. [14:53] so it's not entering the proper creds? [14:54] and why does this FF look so funny? It's like it's running in Windows 95 :D [14:55] hatch: i am around, i'm about to be on a call. talk after that? [14:58] hatch: it shouldn't ask at all [14:59] jcsackett: ok np, I was just going to say that I needed to add scrollview-base-ie to make your carousel stuff pass it's tests in IE - so I Just wanted to update you on that change [15:00] hatch: ok. i'm working with TabView now; i'll check if we need similar things for that. [15:01] bcsaller_: so viewing the chrome tests I see it doesn't ask in Chrome, only in FF? [15:02] hatch: that appears to be the case, there is now a different issue in Chrome showing up in the same test (which appears very similar to what we were seeing in IE (which I find oddly encouraging)) [15:03] oh jeeze - if you run on EC2 do they all pass? [15:04] just wondering if it's maybe a canonistack issue [15:04] not sure how it would be.... [15:05] but who knows [15:05] rick_h_ so what did you need me to look at? [15:05] hatch: nvm, got it. new Y.Node !== Y.Node.create :( [15:05] out yesterday is giving me a case of the mondays [15:06] ohh :) [15:07] hatch: haven't tried EC2 in a while [15:09] rick_h_ will the charmstore stuff be under /bws as well? [15:10] or is that just for these custom views? [15:10] hatch: ok, so for the moment it's there to 'feature flag' the work in progress. [15:10] oh ok so at some point the bws will go [15:10] ? [15:10] hatch: I think at some point closer to release we'll chat on making sure urls are generated correctly and drop them [15:11] gotcha [15:11] hatch: rgr, from the looks of things it should be a simple one-day'er to drop/update those [15:11] ok I'm working on the routing bug from the sprint and making up url stories [15:11] and for now we want them hidden so keeping it around [15:11] ah, well to help QA things we want them to work under bws/ I think [15:12] but we can work on something else if that causes you more work/bugs on your end [15:12] nope that's fine it's just a path [15:12] but in making the stories it didn't make sense to keep them [15:13] so I'm glad to see they will probably be going :) [15:13] yea, add the "hide in progress work from people using the charm for production" story :) [15:13] lol [15:15] gary_poster: do you have a minute for a quick hangout? [15:17] sure frankban [15:18] gary_poster: thanks, I am juju gui hangout [15:18] *I am in the [15:37] bcsaller_, hatch, whoa! "Jenkins is back to normal" in some better universe where passing tests is normal! yay! :-) [15:39] gary_poster: umm... [15:39] lol - well the unit tests all pass [15:41] I can't make sense of this new world but I feel like I should enter some kind of lottery very soon [15:41] hey Chromebooks will FINALLY be available in Canada [15:41] bcsaller_: lol dooo it! [15:45] bcsaller_: did we decide that /notifications/ should be under :gui: ? [15:46] hatch: I don't recall that, there are two views through, the persistent one on every page and the view all one. View All won't have other namespaces active if the design stays the same and the persistent view will call next() anyway [15:47] alright and we decided that show_environment will be on * [15:47] right? [15:49] hatch: there is some drift between the designs of today and tomorrow. If by * you just mean an open match at the end we can make a case for that, but its not present on pages like /service/x and /charm/x now [15:53] hmm I updated chrome and now I get no console logs... [16:14] hi rogpeppe: if you have time, could you please review https://codereview.appspot.com/7598043/ ? [16:14] frankban: looking [16:15] rogpeppe: thanks [16:39] bcsaller_: do you know where the event listener for the juju logo is? [16:40] hatch: think its just an [16:40] oh woops so it is === teknico_ is now known as teknico [16:40] :) [16:41] looks like it's still being captured by the app [16:42] anyways...the issue with it is that it's supposed to drop back to 'root' but what IS root supposed to be :) [16:43] I would say the environment [16:45] but if you're on juju:8888/:gui:/service/memcached/ and click that button it stays there [16:45] because it's not editing the namespaced paths [16:47] hatch: the navigate event handler would catch all the 's [16:48] yeah - I'm just talking [16:48] :) [16:48] I think I'll create a route on / to show the environment view [17:12] bcsaller_: I know you need a break from your CI stuff ;) Can you take a look at http://bazaar.launchpad.net/~hatch/juju-gui/ns-routing/revision/436 and let me know what you think of my approach [17:13] hatch: a welcome chance to do something else [17:14] feel free to tell me it's wrong ;) but then you better have a better way :P [17:14] bcsaller_, figure out some way to wrap up what you have done and we'll do some kind of handoff to someone tomorrow [17:14] someone may be me :-P we'll see [17:15] bcsaller_, if "wrap up" means "we pair for a little bit and you show me what you've learned" so be it [17:16] gary_poster: I'll do one more round of tests and then we can talk if you have time [17:18] bcsaller_, ok. Would prefer to wait till tomorrow when hopefully I will be more rested. Happy to let you have a break though. If you have a small slack task in mind (including reading up on some at-least-slightly relevant technology :-) ), fine, or we can talk through something. [17:18] gary_poster: tomorrow is fine [17:18] cool thanks === deryck is now known as deryck[lunch] [17:28] bcsaller_: so what do you think? [17:30] hatch: we have code like that in check_credentials, so its already ok, I'd refactor to use the one call in both places for now [17:31] hatch: and I'd prefer using the url() method of router to build the url in service [17:32] oh right - I agree [17:32] and I'll look into the check credentials fn [17:33] thanks [17:45] hatch: got time for that chat now? [17:50] yup, see u in guichat? [17:51] hatch: yep [18:10] frankban: you have a review [18:10] thanks rogpeppe! [18:11] frankban: np. i'll be really glad to see that branch land! [18:11] frankban: i'm off now [18:11] frankban: g'night [18:11] night rogpeppe [18:12] rogpeppe: good night [18:21] rick_h_: camelCase..... :P [18:21] * hatch ducks [18:21] yea I will. [18:21] that's cleanup work [18:23] hmm and I can't see anything in your code that stands out as 'will break other code' [18:26] yea, wtf...oh well keep poking it with a stick [18:27] ok I'm going to take lunch, when I get back I'll revert my env back to 0.8.x and then run those tests again [18:40] hatch: ok, in looking the exception is through the scrollview-paginator-debug. So I think maybe what's going on is I'm not cleaning up after myself in my tests that use a slider widget on it === deryck[lunch] is now known as deryck [19:07] hatch: found it! thanks for taking a look that I wasn't insane. [19:09] haha what was it? [19:13] hatch: so the slider doesn't like not having any items in its list [19:13] ohh, no it doesn't [19:13] :) [19:13] so in testing, it would die trying to scroll to with no first index item and cause issues [19:13] woops [19:14] since I gave it no data to test with [19:14] yea...will have to chat with jcsackett about seeing what we can do to make it 0-item safe [19:14] pita for testing to know that you need to popular a widget buried in a view [19:14] popular/populate [19:32] bcsaller_: the url method on _Router is only available in the parent app.js does the service view have a reference somewhere to that app? [19:33] hatch: no, we push deps down from the app, we don't just bind the top level object [20:04] hi gary_poster, have a moment to talk about get_endpoints? === BradCrittenden is now known as bac [20:04] sure bac [20:05] guichat? [20:05] bac, yes it is free [20:17] gary_poster: can I delete some tasks which no longer apply on the kanban? [20:18] hatch, yes. If they are linked to bugs please mark them is won't fix or invalid or fix released as appropriate [20:18] sounds good [20:18] *as [20:20] rick_h_ do you want this ticket assigned to you or is it no longer required? https://bugs.launchpad.net/juju-gui/+bug/1130791 [20:20] <_mup_> Bug #1130791: Add open/close api to Y.SubApp < https://launchpad.net/bugs/1130791 > [20:21] hatch: I think no longer required [20:21] alright - I remember we chatted about it, but I wasn't sure if the api would still be required in general or not [20:49] * bac -> dog walk, ship spotting [20:54] * Makyo dogwalkinate. [21:02] bcsaller_: in /service/:id you can click on the units, what handles that event? I can't seem to find it [21:02] ugh [21:02] I just did [21:02] *facepalm* [21:15] I Just ran into this issue with my namespace work https://bugs.launchpad.net/juju-gui/+bug/1157199 [21:15] <_mup_> Bug #1157199: Invalid charm link when viewing a unit < https://launchpad.net/bugs/1157199 > [21:15] so I need to fix it first [21:27] can someone who is running node 0.8.22 tell me which version of npm they are using? [21:28] jujugui^ [21:28] node --version; npm --version [21:30] hatch: v0.8.22 [21:30] 1.2.14 [21:30] thanks [21:31] I just had some package issues when reverting so I wanted to make sure I picked versions that were working for others [21:32] I am really starting to like tmux [21:32] it's just so easy [21:36] wrt bug #1157199 where is that link 'supposed' to go to? [21:36] <_mup_> Bug #1157199: Invalid charm link when viewing a unit < https://launchpad.net/bugs/1157199 > [21:37] the code is attempting to pull the json from jujucharms which isn't allowing the CORS request [21:40] hatch: isn't it http://uistage.jujucharms.com:8080/charms/charms/precise/mediawiki-3/json/ being generated improperly? [21:41] yes the url is wrong, and I've fixed that - but it requires the CORS request to go through properly but jujucharms.com is rejecting it [21:41] at the service level the charm link is generated properly and seems to work [21:41] ok let me try again [21:43] oh doh, I needed to regen the templates [21:43] heh [21:47] fixed [21:47] :) [22:02] If anyone wants some easy code reviewing before they head off for the day https://codereview.appspot.com/7546053/ 8ln diff :) [22:09] jujugui i'm going to modify the staging server to use an environment with a thousand units and increase the refresh cycle on it to 30m [22:09] zomg! [22:09] hatch: ok [22:12] hatch: done [22:12] thanks sir [22:13] hazmat: I've always wondered, is this actually creating these instances or is it just a simulation? [22:13] hatch, simulator [22:13] hatch, otherwise gui dev would be alot more expensive ;-) [22:13] haha understood [22:14] is it even possible to run multiple instances on one machine? [22:27] hazmat, sounds good [22:28] gary_poster, i'm trying it out right now.. still in progress.. it takes about 10m to create a set that large in the simulator.. [22:28] I bet [22:29] hatch, yes.. lxc or virtual-maas [22:29] hatch, or if you met multiple simulators yes.. that's possible to. [22:29] cool - I could see that being helpful for people who want to setup a small blog with caching [22:29] or simple webserver etc [22:35] gary_poster, re stage server.. its about 15m for the simulator.. to keep this scale present, it would need to divorce the gui updates from the content reset, continue the former every 15m, the later maybe 6hrs.. [22:36] hmm.. maybe 20m [22:36] ugh.. maybe i should just give up the ghost on this.. [22:39] that would be pretty cool though [22:39] i got it working locally but didn't time the simulator all the way through [22:40] the gui was fast though [22:42] at that level I'd be concerned about memory creep [22:42] but that's for another time :) [22:43] aha.. its active.. [22:43] http://uistage.jujucharms.com:8080/ [22:44] lol [22:44] yeah don't leave that open too long ;) [22:45] it's crashing my tab [22:45] but it loaded the first time [22:45] ok it doesn't crash if the inspector isn't open [22:46] hazmat: is this for a demo? [22:46] hatch, no just trying to show the gui is usable at this size [22:46] for discussion [22:47] ohh - well it's gone up 10MB since I opened it [22:47] lol [22:47] we have some work to do ;)\ [22:47] although evidence of the opposite is also progress [22:48] hmm there is some oddness here [22:48] hatch, i only show about 65k of mem usage [22:48] less than my gmail tabs [22:49] the application seems rather stable in the heap shots but the realtime inspector is steadily increasing [22:49] 65k? it's using ~12.4MB :) [22:49] interesting for me memory usage is decreased without active focus [22:49] down to 49k [22:49] hatch, which browser are you using? [22:49] that's normal [22:49] Chrome [22:50] you can't use the system inspector to view the resource usage of tabs [22:50] because it offloads things to different processes [22:51] in the dev tools click the 'Profiles' tab [22:51] then take heap snapshot [22:52] that will give you the memory usage of the page [22:52] then you can do things, and take another heap to see if it goes up [22:54] hatch, i'm using the chrome task mangaer [22:54] hatch, tools like profiles perturb the observed behavior [22:55] hatch, ie they account for a significant memory consumption by themselves [22:56] task manager is showing me 141MB usage for the tab vs 12.4 in the heap snapshot :) [22:58] hatch, it would be nice to verify those numbers without any profiling [22:59] which can account for significant memory by itself [22:59] bcsaller_: looks like my changes to the routing code has also fixed the history navigation issues we were seeing at the sprint [22:59] hatch: interesting, be nice to see what you had to change [23:01] bcsaller_: I'm proposing it as a WIP soon [23:02] hazmat: I've always used the heap snapshots in the past to monitor memory usage over an apps lifetime [23:02] and it's served me well [23:02] ack [23:03] it might not be 100% but it's at least consistent haha [23:03] hatch, we're seeing such *wildly* different numbers ... i went to look for causes [23:05] when you took a heap snapshot did you not get ~12MB? [23:06] i do [23:06] oh ok perfect then - I think the task manager takes into consideration the whole tab instance on the OS [23:07] but as far as your 65K - I don't know about that, a photo is larger than 65K ;) [23:08] hmm.. maybe that's 65k *k [23:08] oh maybe [23:08] I'm in OSX also [23:08] but os process manager shows 121mb for the chrome in total [23:08] so there might be some difference there [23:09] oh ok so mine showed ~140MB so we are pretty close on that [23:09] assuming that the task manager and process manager are close [23:09] s/close/reporting the same thing [23:10] yeah.. that seems more reasonable [23:10] * hazmat tries the staging site on a phone [23:11] hehe I was just loading it on an N7 [23:12] hatch, chrome or android browser? [23:12] * hazmat is trying safari 6 [23:12] chrome [23:13] it loads and displays the UI but it doesn't appear that I can click on anything [23:13] yeah the SVG elements aren't triggering any of the events [23:13] but the top navigation is functional [23:13] hatch, can you drag elements? [23:13] the bottom navigation isnt' on the screen [23:14] negative - but I can pan the environment [23:14] oh wait [23:14] two finger drag [23:14] the elements [23:14] but twofinger tap doesn't trigger the click [23:15] bcsaller_: doesn't look like I'll be getting this branch proposed...the tests are failing and proving to be difficult to solve [23:16] hatch: I can still glance at the WIP if you want [23:16] just push and give me a link :) up to you [23:16] https://code.launchpad.net/~hatch/juju-gui/ns-routing [23:17] sure - I was going to propose so it would give you a proper diff [23:17] unless you can somehow diff trunk easily in LP [23:21] hazmat: looks like it needs a little work to get it working on mobile but shouldn't be too much [23:21] how was it on safari? [23:22] hatch, not so hot... chrome a little bit better, never got to the rendered graph in either but gave up after 3m [23:22] hmm - was that on an ipad? [23:22] or iphone? [23:22] iphone5 [23:22] ahh - I'll try FF on my old android phone [23:23] I don't have high hopes heh [23:23] i was demoing the normal staging gui on the phone earlier it worked reasonably well [23:23] rendered, drag, pan and zoom [23:24] oh ok - so the proc/ram increase must have put it into limp mode [23:25] ok FF on Android 2.2 works [23:25] I can even get the popups [23:25] actually it looks like I can do everything [23:26] it appears to be working as a desktop browser [23:26] it's even reasonably fast ui rendering [23:26] but it gave me a 'script taking too long' error at first [23:38] k, downsizing to the 100unit env [23:39] and keeping 30m refresh/reset [23:40] that's pretty cool how well it works at that level without any real work done to improve performance [23:43] interesting.. the heap delta from 1000k to 100 is from 11.8mb to 11.0-10.4mb (both fresh) [23:44] almost the same 11.8mb on one.. [23:44] hatch, can you confirm re heap size on current uistage site [23:44] sure one second [23:45] 10.9MB on raw load [23:46] so it looks like it's not the ram causing issues but the adding processing of 900 extra units [23:47] that can be remedied however