[12:33] hey rick_h_, let's talk about bug 1287977 when you get a chance [12:33] <_mup_> Bug #1287977: searching for rabbitmq locks the GUI up until all 400 results are loaded and rendered [12:33] bac: sure thing, let me grab my headphones [12:34] bac: https://plus.google.com/hangouts/_/7ecpjmmfgnbo8cqqqaqelr0j90?hl=en [13:02] rick_h_: confirmed my rolled days have been added back. thanks. #shouldnotbethishard [13:03] bac: cool, thanks for checking up and glas we got it squared away [13:07] * bac reboots [13:17] * frankban lunches [13:32] * bac trusty vm is being uncooperative [13:34] bac: did you get trusty to run charmworld? I hit PPA issues when I tried. [13:34] rick_h_: yeah, i can run cw...when i can log in [13:34] gotcha [13:34] right now after typing my password into the ui login it locks up. [13:35] can get in via ssh [13:35] cool, will be curious to hear what you used for the ppas in the build when it's running better. [13:35] would that be unity-greeter? [13:35] bac: oh! sec. I saw someone post asking for a vmware fusion help with X issues (well unity?) [13:35] rick_h_: was that barry yesterday? his issue was crashes under weird circumstances [13:35] bac: https://twitter.com/pumpichank [13:35] oh, ok [13:36] but he'll be the first person i bug... [13:36] * rick_h_ tries it out with non-unity [13:36] wait, how's that? [13:37] I run Awesome, curious to see if it hits the same issue [13:37] I can get in now, pulling the latest updates and will try again [13:38] this is probably ungood [13:38] [ 15.985841] type=1400 audit(1394199451.466:74): apparmor="DENIED" operation="open" profile="/usr/lib/telepathy/mission-control-5" name="/var/lib/lightdm/.config/libaccounts-glib/accounts.db" pid=2352 comm="mission-control" requested_mask="rw" denied_mask="rw" fsuid=110 ouid=110 [13:39] yea [13:40] there was an update for unity-greeter that was "held back" from apt-get update. manually installing it. may help, may fry everything. [13:40] i've never understoon the 'held back' stuff [13:41] s/understoon/understood/ [13:42] whoo hoo [13:42] not only does it work, the desktop is now watermarked '14.04 LTS' -- that'd been bugging me! [13:43] hah cool [13:46] rick_h_: so i'm back. if you want to talk about charmworld ppa's let me know. [13:46] bac: not right now, I'll bug you about it later [13:46] good answer. :) [14:15] rick_h_ yeah two qa's would be fine [14:16] hatch: cool, I'm working on one right now [14:33] benji: any idea why matt's cw branch didn't land? https://code.launchpad.net/~makyo/charmworld/charm-urls/+merge/209556 [14:33] * benji looks [14:34] i see nothing there. will look at jenkins [14:35] bac: http://162.213.35.28:8080/job/charmworld-autoland-lxc/313/console [14:35] jenkins is not having a good day? failed autolanding permission issue? [14:36] rick_h_: and then it just gave up this morning? [14:36] bac: well looks like it's been running over and over [14:36] http://162.213.35.28:8080/job/charmworld-autoland-lxc/ [14:36] yeah but stopped at 03:38Z [14:36] about every 10min [14:37] oh hmm [14:37] nothing in 9 hours [14:37] yea guess so. [14:37] maybe that's built in [14:37] abentley: ^ any insight on the lander there? it's very unhappy on the autoland script [14:37] wonder if something got changed recently [14:37] rick_h_: otp [14:37] abentley: rgr, thanks [14:43] rick_h_ so I don't get held up I'm going to land this branch and we can get a few deep qa's after it's all done [14:43] hatch: rgr [14:44] only one more to go [14:44] *squeeee* [14:45] of course it's the biggest and most complicated lol [14:45] saved the best for last! [14:45] oh wait no there are two left [14:45] service relation and service overview [14:46] relation should be easy [14:46] famous last words! [14:46] hah [14:48] rick_h_: the MP did not have an approved revid. i've set it now. can't get a build to kick off, though. [14:49] abentley: jenkins for charmworld gave up and stopped processing jobs about 11 hours ago. ideas on how to get it going again? [14:49] bac: otp [14:49] np abentley. ping me when you've got a sec. [14:50] bac: I just added an approval since one was pending. I think that'll get the bot to land the branch. [14:50] benji: it looks like it got angry with us and now is on strike [14:51] heh [14:51] this should let it land once the scabs are bussed in [14:56] apparently you have to be careful what you search for on G+ [14:56] hatch: s/G+/internet [14:56] lol - I was definitely not expecting NSFW photos when searching 'black pug' [14:57] benji: i manually kicked off a build. [14:57] k [14:57] rick_h_: how comes the coming soon simulator doesn't show the landscape restart or security updates on the service block? [14:58] luca: ummm, I don't know. Should they? /me isn't up on the latest in landscape integration UI design. [14:58] rick_h_, bac: pong [14:59] abentley: i figured out how to manually kick off a jenkins build. will follow up if it fails again. [14:59] It looks like it's trying to merge charmworld into itself. Also that it's trying to merge in a revision that does not exist. [14:59] abentley: 314 was my mistake [14:59] bac: Okay, I'll leave it with you for now. [15:00] rick_h_: I'm not sure, last I knew the restart and sec update should show on the service block, can anyone confirm if they show in the real environment? [15:04] frankban: I think I've taken a wrong turn on this branch, do you have a minute to chat? [15:04] luca: so they show in the inspector with links for upgrade and reboot in the unit list [15:04] guihelp: I need two reviews (no QA) for https://github.com/juju/juju-gui/pull/167 . anyone available? thanks! [15:05] luca: I don't see anything on the service block [15:05] frankban on it [15:05] frankban: will look in a sec [15:05] benji: sure, give me a minute (coffee) [15:05] luca: so I'm not sure about the block level. It seems it got moved to units? [15:05] hatch, rick_h_ thanks [15:06] rick_h_: last I knew it was in both places, no worries if it's moved :) [15:06] luca: yea trying to glance through the code quickl [15:13] benji: hangout? [15:14] frankban: sure, I'll make one [15:16] frankban: https://plus.google.com/hangouts/_/7acpiae8apl54n76eg0e0mr1ak?hl=en [15:28] hatch: interesting: is passing context to map more js idiomatic than self = this? [15:29] rick_h_: also, there are no hover states for the inspector [15:32] frankban so the difference is because the compiler has to keep the outer scope in memory because it's referencing that variable. If you pass it in, internally it binds the method to that context so it can properly clean up the wrapping method [15:33] hatch: makes sense [15:33] it's a pretty trivial performance gain hah, but it cleans the code up a bit imho [15:33] the compiler can then 'compile' the callback fn as well which can lead to some really big improvements in a loop [15:36] luca: k, sounds like perfect bugs to file for a new starter to hack on [15:37] rick_h_: I'll file away [15:38] benji: sadness http://162.213.35.28:8080/job/charmworld-autoland-lxc/315/console [15:40] "npm ERR! Error: UNKNOWN, unknown error [15:40] that's an odd error [15:42] benji: hmm, wonder if that's old npm issue [15:42] benji: we had to update npm in the ppa because of npm no londer allows downloading on the self-signed certs [15:43] forgot charmworld used npm at all [15:43] the fact that it is the very first npm install in the run (as far as I can tell from scanning the output) suggests it might be [15:43] benji: yea, when we got it on the gui side we had a more helpful error [15:43] benji: but it's a know recent change [15:44] known that is [15:56] jujugui meeting in 4 [15:57] benji: my irc client dropped out. did you have ideas on the jenkins failure? [15:57] bac: Rick and I think npm's recent global breakage because of the cert re-issue is a good first thing to check [15:58] ...especially because the failure is the first time in the run that an npm package is installed [15:59] benji: ok, perhaps we can chat after the meeting [15:59] k [16:00] oh, meeting [16:03] jujugui: review of charm-proof - https://codereview.appspot.com/72170043 [16:06] jujugui call in 9 [16:06] jujugui call in 3 [16:07] whoa [16:07] jujugui call in 10 [16:07] irc has gone crazy for me. just getting a lot of queued messages [16:07] jujugui call now [16:07] Same [16:09] woah that was a lag lol [16:11] someone is probably DDOSing freenode again [16:13] Makyo: hatch frankban benji bac rick_h_ https://docs.google.com/a/canonical.com/file/d/0B7XG_QBXNwY1SjJ5anNYblJva1E/edit [16:13] Makyo: hatch frankban benji bac rick_h_ https://docs.google.com/a/canonical.com/file/d/0B7XG_QBXNwY1N2NHXzI5d0NBZ2M/edit [16:14] jujugui lf a really quick review/qa https://github.com/juju/juju-gui/pull/168 [16:14] luca fyi `jujugui` pings the whole channel :) [16:14] hatch: oh [16:14] well - we all have it set up to do so [16:14] jujugui awesome [16:14] did it work? [16:14] :) [16:14] yep [16:15] I see [16:15] ugh more damn inspector design changes [16:15] lol [16:15] lol [16:15] I feel you [16:15] haha [16:15] I kind of wish the swooshes were part of the design [16:16] * hatch I like the new design though...looks 'tighter' [16:16] thanks :) [16:16] and for whatever reason I love the deploy bar [16:16] lol [16:17] haha [16:17] it might change visually, spencer is working on the hard task of actually making it look good [16:17] really? I think it looks great now haha [16:17] this is why I'm not a designer I suppose :) [16:17] ha [16:18] most probably :P [16:18] hehe [16:18] hatch: looking at your branch [16:18] frankban thanks! [16:19] bac: https://codereview.appspot.com/72170043 returns "No issue exists with that id (72170043)" [16:19] luca if I could....I'd like something which maybe pointed the 'configuration' button out to the user in the first panel [16:20] Just so they wan't like 'where do I configure this?' [16:20] oh man I can't type [16:20] just so they don't say 'where do I configure this?' [16:20] :) [16:20] or we could open it by default [16:20] that would probably be the best [16:20] has Anyone Really Been Far Even as Decided to Use Even Go Want to do Look More Like? [16:21] lol [16:21] I'm actually laughing [16:21] ohh [16:21] :) [16:21] English is hard [16:21] haha [16:21] when a user deploys a charm from the store [16:21] the config panel will be open [16:22] +1 [16:22] they won't see the status panel unless they navigate to it or close and open the service [16:22] awesome [16:22] but we want to add a little on boarding to the status panel to clue people in on what happens once they deploy, since some people may never deploy [16:22] yep that's a good idea [16:29] frankban sorry I totally forgot to do something in that branch [16:30] hatch: also have you seen the selenium failure? [16:30] yeah I did [16:30] I wonder why it's crashing so often [16:32] oh that's why I missed it....it's not even in trunk [16:32] heh [16:35] rick_h_ regression in master https://bugs.launchpad.net/juju-gui/+bug/1289469 [16:35] <_mup_> Bug #1289469: Remove relation button missing from relations tab in inspector [16:36] hatch: k [16:36] rick_h_ maybe what we need to do is create a master list of functionality and interactions so before we do a release we can check each one as well as doing the exploratory qa [16:42] hatch: ugh, qa checklist... hatch well let's start by adding a test [16:42] it's the righ thing, bug escaped, write a test, fix it [16:43] rick_h_ well no it's not really a qa checklist it's more like 'these are the things that our application does' does it still do that [16:43] I mean...yes totally need tests :) [16:44] but this is more of a higher overview [16:44] hatch: right, so the thing to look at is qa vs tests. A lot of qa we can automate with functional tests [16:44] vs a checklist to run down [16:44] I do think we need to do more functional tests, but this one seems like just a bug [16:44] something we can test for [16:44] I think as we modularize things more then the tests will become a lot easier to write as well [16:48] hatch: definitely [16:52] hatch: frankban they released some new stuff with a new sauce connect and such in the last week. I wonder if we're hitting something in their upgrade [16:52] hmm well I hope not [16:52] :) [16:53] yea, not sure. [17:10] sorry for the delay frankban [17:10] frankban: +1 thanks for the machine view updates there. [17:10] rick_h_: thank you! [17:15] * rick_h_ runs for some grub [17:35] so confused... [17:35] look for the turtles [17:35] they'll lead you all the way down [17:36] we have events here which don't attach to anything or have callbacks [17:36] lol [17:36] how the heck didn't these throw errorsx [17:36] haha [17:52] jcsackett: got a question about the queue and charmworld api [17:53] jcsackett: as in, is the queue exposed int he charmworld api [17:53] queue being review-queue [17:53] marcoceppi: i do not believe so, no. [17:53] jcsackett: how hard would it be to have it added to the api? [17:53] and who would I have to bug to get that worked lined up? [17:54] marcoceppi: i don't know off the top of my head; i haven't touched API code in some time. [17:54] * marcoceppi bugs rick_h_ ^ [17:54] marcoceppi: to what end? [17:55] jcsackett: I want to be able to get the charm review items (not the proof or au questions) as an api endpoint. So created, last updated, status, and link to review item [17:55] marcoceppi: i figured that; i'm just curious what you want to consume said endpoint. [17:56] marcoceppi: hmm, not sure at the moment. You can file the bug and we can see. We'd be happy to help mentor/review some code that did what you guys need :( [17:56] :) [17:56] that is [17:57] nananananananananana bug man....submitting bug man [17:57] jcsackett: so, to start there's a charm review command in charm-tools that poorly re-implements the lp code in the review queue, then also some automation bits for the juju solutions team that reads data from teh queue [17:58] rick_h_: I was hoping it would be a low level of effort, I'm happy to hack on it since it really only helps me [17:58] marcoceppi: ah, dig. [17:59] marcoceppi: ok, not a problem. We're currently under a crunch to get stuff ready for bundles announcement next week but can try to help look at it after that. [17:59] frankban: (if you're still around) sorry, that RV got deleted due to chunk mismatch. MP at https://code.launchpad.net/~bac/charm-tools/bundle-proof-warnings/+merge/209754 [17:59] rick_h_: cool [18:00] marcoceppi: could you have a look at this extension to charm-proof? https://code.launchpad.net/~bac/charm-tools/bundle-proof-warnings/+merge/209754 [18:00] bac: can do, I need to finish this review for the same bundle announcment but after that (say, 30 mins) I can take a look [18:00] marcoceppi: cool. will take you just a jiffy. [18:01] bac: actually, this lgtm, I'll pull and make test to verify [18:01] bac: there will likely be a 1.2.10 release tonight of charm-tools that has this and the other fixes you submitted earlier [18:01] jujugui review/QA for minor branch: https://github.com/juju/juju-gui/pull/169 (QA: drag and drop a charm with inspector open, ensure that there are no 'undefined' errors) [18:01] marcoceppi: thanks. note that the version of charmworld to generate those warnings isn't deployed yet, but proof doesn't care [18:02] bac: right, cool, thanks for getting this in before that's live o/ appreciate it [18:02] Makyo: woot will look [18:06] what was the documentation make command to clear out the 'non documented' functions? [18:06] since they are now documented [18:06] make undocumented? [18:06] will try [18:07] Yeah [18:09] 15 test failures for the overview switchover [18:09] not baaad [18:11] Makyo: thanks for that, not sure I would have seen that. The traceback was off a different area if I recall [18:11] Makyo: that's one big method, maybe we can break it up some to help down the road [18:12] rick_h_, agreed. Everything was 'working' in our code, it was d3 trying to assign attributes with undefined values through a callback. [18:13] Makyo: yea, I mean the UI worked just dumped errors. So it was a strange one to start with [18:15] my keureg machine broke :'-( [18:16] good, get some real coffee :P [18:16] but then I have to leave the house [18:16] IT"S COLD OUT THERE! [18:16] http://www.amazon.com/Primula-Aluminum-Stovetop-Espresso-Coffee/dp/B001J1L59E/ref=sr_1_2?ie=UTF8&qid=1394216192&sr=8-2&keywords=moka+pot [18:16] still make it at home [18:16] oh I'm not an espresso fan [18:16] drip or nothing [18:17] :) [18:17] ugh [18:17] :D [18:18] 9 failures..... [18:18] oo this one hasn't gone back up after I've fixed ones [18:18] this is unprecedented I'm not sure what to do...lol [18:23] rick_h_, I see that my bug was indeed a duplicate. Sorry about that. I want to assure you I did search before I created it. [18:23] mbruzek_: it's all good [18:24] mbruzek_: we appreciate the feedback. I just want to keep the discussions together for ease of everyone working on things [18:24] Yeah that makes sense, I wish I would have found it in my search. [18:24] don't worry about it. Can't know about all the bugs in all the projects and sometimes search-fu fails [18:24] appreciate you filing [18:24] Really? [18:25] I would imagine this issue has been talked about before. I hope it did not reopen a long dead discussion. [18:25] yea, and deisgn is aware of it. I think most of us agree with you. The thing was that sometimes design is pulling standards from other places. [18:26] phone design, other website, and it takes some tweaking to get right [18:26] more people saying it's confusing helps us look into it more [18:26] Sure and on the bug I opened I figured you already had a discussion and made a decision but I just didn't know where it was written down [18:26] rick_h_, understood. [18:27] well in that bug luca mentions that they're working on it [18:27] so until they figure it out and do some testing/etc we're running with what we have [18:27] Ok [18:27] Have a good weekend rick_h_ [18:27] you too [18:27] http://i.imgur.com/upi6iwF.png [18:27] bahahaha [18:31] jujugui lf a review on https://github.com/juju/juju-gui/pull/170 the final viewlet to view diff [18:37] rick_h_: i think i know what is going on with search [18:38] our input terms are tokenized. so 'rabbitmq-server' gets split into parts and we return everything with any reference to 'server' [18:39] rick_h_: http://paste.ubuntu.com/7051463/ [18:43] bac: oooohhhhhhhhh [18:44] do we really want stemming? [18:44] bac: wow, ok. What about the revisions ? [18:44] bac: well it was something requested yes. because of things like this. rabbitmq [18:44] don't think revisions is an issue [18:44] bac: the dual cinder int he results for rev 17 and 19? [18:45] it shouldn't be returning the server ones unless they also match rabbit [18:46] rick_h_: here are the charms and bundles returned by a search for 'rabbitmq-server' http://paste.ubuntu.com/7051502/ [18:46] hatch: hmmm, so tokenize but treat them as an AND vs an OR? [18:46] bac looking [18:46] rick_h_ yeah [18:46] bac: right, so line 40/41 [18:47] so, yeah, cinder returns multi versions of the official charm. other packages do not. i missed that. [18:47] rick_h_ but it also needs the ability to backup....say if there are no matches for 'rabbitmq-server' but there is for 'rabbit' then it should show those (imho) [18:47] yea, it's not going to help us a ton [18:47] the revisions stuff [18:48] bac: do we get weighted scores back from search? [18:48] bac: can we tweak the threshhold on something to help cut out some of the noise? [18:48] yes [18:48] or maybe there's a field we're weighing too highly [18:48] to get back so many 'server' results [18:49] we can't know when we tokenize which of x,y,z is creating noise [18:49] maybe we can just use the google search thingy [18:49] :P [18:49] bac it should query on 'rabbitmq' 'server' 'rabbitmq-server' [18:49] but i don't think this is due to a recent change. [18:49] bac: what do you think of checking on the revision issue (that seems like a bug bug) and maybe there's some suggestion for search we can add to help with the situation [18:49] then we should be able to weight them properly [18:50] bac: yes, it seems it's not directly a regression but getting more charms and search working too well :) [18:50] hatch: full string gets more weight but you can't differentiate between the others [18:50] hatch: ES is doing that tokenizing, not us [18:50] we should be able to see if the scores cut off nicely at some point though [18:50] hmm I've never worked with ES....last project was with Solar and we were able to do what i mentioned [18:51] Is this a limitation of ES? because ES and Solar are both based on Lucene no? [18:51] and maybe adjust the weight of description or something that's catching so much 'server' [18:51] ES is lucene based [18:51] hatch: solr and it depends on what you do. Right now, the tokenizing/scoring is done in ES land [18:52] hmm - I don't know enough about ES to say, but it sounds like we 'should' be able to do this [18:52] would kind of make for a crappy search tool if you couldn't tokenize and weight heh [18:52] yes, i should hope we should [18:53] I'm sure there is some 800 page book you could read on it bac ;) [18:53] i'll dig into the ES docs [18:53] heh yea [18:53] bac: ok, so you're ok moving forward then? [18:53] there are too many 800 page books [18:53] haha :) [18:53] this industry needs more 80 page books [18:54] i remember looking through the index of a 1000+ page book on HTML or CSS or something and there was reference to the Dining Philosophers. stupid editors. [18:55] moar pages = better right? :P [18:55] must be [18:56] rick_h_ have a second for a quick hangout? [18:56] sure [18:56] https://plus.google.com/hangouts/_/76cpisv6qllekpbpm7ialobo28?hl=en [19:03] bac and hatch: the _get_text_query method in charmworld/search.py generates queries with weights for the different fields; that might be a good place to start [19:04] Oh I don't really know enough about the ES functionality to contribute, I was just commenting on what we 'should' be able to do coming from my experience on a project with solr [19:05] hatch: it sounds like you are now our ES expert [19:05] hahahaha [19:05] woot! [19:06] I'll add that on top of my Python and Go training [19:06] NO BIG DEAL [19:06] jujugui I still need a review on https://github.com/juju/juju-gui/pull/170 :) [19:06] hatch: working on it [19:06] awesome thanks [19:07] it's in one of these tabs here [19:26] it so much nicer that lxc containers come up in the same IP address each time. this VM keeps changing it's ip [19:27] the vagrant one keeps it's ip, my lxc's on my juju box changes the ips every time [19:30] I'm going to go grab some lunch [19:35] Alright, that branch merged. Headed out. Have a good weekend, see you all Tuesday. [19:35] Makyo: have a good weekend! [20:14] so I suppose I Should keep moving on this inspector business [20:17] hatch: :) [20:36] rick_h_ are you going to need me for UDS next week? [20:36] or are you handling it? [20:36] I just need to know if I should go get a spray tan and stuff..... [20:37] bahahaha [20:37] hatch: well I wanted to see if I could get a volunteer [20:37] hatch: I proposed we do a bundle/quickstart demo session [20:37] and answer any questions and such there [20:37] hmm well I'm stuck in osx land so I'm probably not the best person to demo [20:37] k [20:37] well I'll get someone and if I can't I'll do it [20:37] I can definitely be there to answer questions and the like [20:38] but my goal will be to get someone :) [20:38] I don't think a single proc is enough to stream a hangout :) [20:38] hah [20:43] ok, going to head out for the weekend. everyone have fun [20:43] thanks for all those views this week hatch [20:43] no problem, have a good weekend [21:59] jujugui hey anyone around still? [21:59] hatch: kinda [21:59] what's up? [22:00] oh just wondering if I am going to get a review on the card I'm working on :) [22:00] we put it at a day and it took me 2h so it'll catch me up :D [22:00] hah awesome [22:00] the base class? [22:00] yeppers [22:00] the others will take more time [22:00] ok cool [22:00] but this one is only a few lines heh [22:00] sometimes I love YUI [22:01]