[12:15] ooh, the AttributeError: 'NoneType' object has no attribute 'clone' problem is aparently a symptom of a change to PyPI (they put it behind a CDN) [12:15] hey bac, it looks like your branch *might* have broken jenkins in some new exciting way [12:16] it failed twice in the same way since your branch [12:16] I have no idea why your branch would trigger that error :-/ [12:17] but could you do a bit of investigation and see if you can dupe, maybe by running the tests locally [12:17] ? [12:19] benji: I thought I got that before this. Cool if there's a fix though. [12:19] benji: the thing is that it's loading a package that's locally cached isn't it? [12:19] no fix yet [12:21] I don't think we've set up any local caching. [12:22] benji: the archives directory? [12:22] with closure-linter and selenium? [12:22] I was thinking that path was off when running in the new dir lbox creates [12:23] Maybe. [12:24] benji: ah, nvm. Reading my log it's installing closure_linter, but failing on the dep python-gflags so yea...pypi [12:24] I just installed python-gflags into my system (apt-get install python-gflags) and it is happy now [12:56] morning [13:02] hi hatch [13:03] hey hatch [13:03] bac I talked to you in your absence [13:03] you did not reply [13:03] hey bac, it looks like your branch *might* have broken jenkins in some new exciting way [13:03] it failed twice in the same way since your branch [13:03] I have no idea why your branch would trigger that error :-/ [13:03] but could you do a bit of investigation and see if you can dupe, maybe by running the tests locally [13:03] ? [13:03] gary_poster: ok, will do [13:04] thanks bac [13:34] The PyPI CDN issue has been fixed. [13:38] yay [13:59] luca: jcastro heads up, url change for using the browser on uistage: https://jujugui.wordpress.com/2013/05/29/yesterday-we-landed-a-change-that-moved-the/ [14:01] rick_h_: ok so I need to change the link on the website? [14:01] jcastro: ? which website? [14:01] juju.ubuntu.com [14:01] has a link to ui stage [14:01] jcastro: no, that's just to view it with the browser [14:01] oh, ok, nod [14:01] jcastro: for your own 'how does it look, qa, wtf'ness [14:02] jcastro: vs uistage:/bws/sidebar/... [14:03] rick_h_: I was expecting you to need the link updated, because IS just updated the website like 10 minutes ago. And you know how fate rolls. [14:17] rick_h_: branch is actually updated now: https://codereview.appspot.com/9828043 [14:17] jcsackett: rgr, will look. Thakns [14:17] hatch, Makyo: any chance one of you could take a look at https://codereview.appspot.com/9828043 ? need a second review. [14:18] jcsackett, sure. [14:19] thanks Makyo. [14:22] sinzui: Could you please review https://code.launchpad.net/~abentley/charmworld/shrink-mongo/+merge/165928 ? [14:25] abentley, this is the entire module that does the checking in pocketlint. http://bazaar.launchpad.net/~sinzui/pocket-lint/trunk/view/head:/pocketlint/formatcheck.py [14:25] sinzui: Cool. [14:29] jcsackett: feedback inbound [14:36] anyone up for a second review? https://codereview.appspot.com/9657046/ [14:40] benji: looking [14:40] hatch: is your "Create document about angular vs yui refactoring" card in review because you want people to review it? If so, I have a moment to do so. [14:40] thanks, teknico [14:51] rick_h_: any examples of a docstring for @event with paylod? can't find one. [14:53] jcsackett: yea, hmm. There were examples docs we're supposed to use for the @event stuff. Trying to see if I can find it. [14:55] gary_poster, http://en.wikipedia.org/wiki/Anders_Hejlsberg yeah it was C# [14:55] hatch, :-) cool [14:55] jcsackett: per http://jujugui.wordpress.com/2013/04/26/weekly-update-matt-start-using-yuidoc-event-yuidoc/ since you don't publish the event I'm wrong. [14:55] jcsackett: so nvm I guess. [14:58] rick_h_: ok. [14:59] benji: how about a quick call about your branch? [15:06] jcsackett, FYI, something's up with net, I've been pulling from npm for about 30 mins. Promise I'm reviewing :) [15:06] Makyo: dig. thanks for the update, and sorry for your net woes. :-) [15:10] anyone else seeing giant request times to npmjs.org? https://registry.npmjs.org/uglify-js/1.3.4 for instance is 20s to load for me atm [15:10] * rick_h_ regrets the make clean-all right now [15:11] rick_h_, ^^^ [15:11] Guess it's not just me. [15:11] Makyo: ah, missed the npm part of your comment. [15:11] Makyo: ok, glad it's not just me then [15:12] Makyo: wasn't there work going on to make npm an offline cache for installs? [15:12] Makyo: or did that never get into play. Or just on the charm? [15:13] rick_h_, I thiiink some set up a local cache for themselves a while back, but I don't think it ever became part of the actual process. [15:14] * rick_h_ fears doing a lbox submit ... [15:18] teknico: I just noticed your message. Sure, how about the regular place? [15:19] benji: guichat? ok [15:33] rick_h_, Makyo benji: benji did that [15:33] jcsackett: return review? https://codereview.appspot.com/9841044 hatch or bcsaller as a second please? Ignore the updated generated .json files used for sample data in the tests. [15:33] npm cache [15:33] but for distributions [15:33] in the charm [15:33] not usable for developer [15:33] gary_poster: ah ok, so not in dev. [15:33] rick_h_, ok I can take a look [15:33] right [15:33] hatch: thanks, should make your life easier. [15:34] I hope I don't have to QA that json file :P [15:34] sinzui: abentley though in testnig I'm still getting 1.5-2s requests to /interesting so might be cool to start up some caching or such on m.j.c coming up [15:34] hatch: I said not to :P [15:34] hatch: two .json files generated, just trust me, they're valid ones from m.j.c [15:34] haha [15:34] rick_h_, against m.jc ? [15:35] sinzui: yes [15:35] manage.jujucharms.com [15:35] I like it [15:35] sinzui: that's locally so might not be perfect test case, but noticed it a lot in doing QA [15:35] rick_h_, when I browse m.jc, I often see my requests timeout ...well a timeout between apache and the proxies [15:35] sinzui: and the caching from jcsackett will help as well, but still. 2s for the request and .5s for the transfer was longer than I'd hoped [15:35] sinzui: :( [15:35] rick_h_: How does that compare to requests for static files on m.j.c? [15:36] abentley: looking [15:36] abentley: so each icon is sub 300ms for instance [15:37] abentley: the /interseting OPTIONS request was 130ms while the actual data was 2.3s in this sample load [15:37] that's 130ms including network time [15:37] rick_h_: Gotcha. [15:37] damn. I cannot get m.jc to show me the timeout now. I bet its meager cache is hot right now [15:38] so still all nice and better, but might be nice to stick a card/bug to look for low hanging fruit to optimize [15:38] rick_h_: So it sounds like a genuine perf issue. We could address it with caching, but we might be able to optimize in other ways. I would prefer that, because that would also help users who hit /interesting when it's cold. [15:39] abentley, /interesting is getting a lot of related charms at this moment. Your current card may address that [15:40] abentley: definitely. didn't mean to say we should implement X, just investigate. Once this branch lands we'll see it a lot easier. [15:47] rick_h_, donezo [15:48] hatch: yay thanks [15:48] don't thank me yet! [15:48] well, maybe yay. Depends on what you found [15:48] jk ;) [15:48] lol nothing to serious [15:50] jujugui call in 10; kanban now pls [15:51] hatch: replies inbound. I disagree :/ [15:51] NEVA!!! [15:52] hatch: let me know if you need more convincing :P [15:52] I do, I have a very thick skull [15:52] :P [15:52] haha going to read now [15:52] * rick_h_ goes to get the 'agree with rick curse you' bat [15:55] I think we can reach some middle ground [15:55] investimagatering [15:55] hatch: k, welcome to have a chat post-stand up [15:58] jujgui call in 2 [16:12] adeuring: now that there's a 'promulgated' attribute, I think charmworld.models.is_reviewed should be removed, or redefined in terms of the promulgated attribute. [16:13] adeuring: similarly for reviewed_clause [16:13] in ElasticSearchClient. [16:14] abentley: yes, renaming would make sense. I am not sure about removing the function -- could it happen that we redefine reviewed/promulgated/... again? [16:17] adeuring: It's conceivable that we'd redefine 'reviewed', but I don't think that justifies keeping a one-liner function, especially since I think changing the def is unlikely. [16:18] abentley: well, changing this one-liner helps to change several views at once :) [16:19] I prefer to have this sort of "common definition", even in trivial cases [16:20] rick_h_, replied - lemme know if you would like to discuss [16:23] hatch: guichat? [16:23] yup [16:23] hatch: nvm, new chat [16:23] it's full [16:23] :) [16:24] r u making the new room or am I? :) [16:25] hatch: tried to make one and invite you. You no show [16:25] hatch: https://plus.google.com/hangouts/_/762e94644ca67de98ca216ec502111c4f338a3b2?hl=en [16:30] adeuring: sure, and the main virtue of the "promulgated" attribute is that it can act as that common definition. It would be even better if it were called "reviewed" instead of "promulgated"-- that would make it obvious that we should update its definition if the definition of "reviewed" changes. [16:32] abentley: ummm... what if the meaning of promulgated and reviewed diverges? I chose promulgated as the attribute name because it refects the things done by ptomulgation script [16:35] sinzui, hey. I have two topics. First, I wanted to check with you as to where we are with m.j.c stability. We had said that it would be ready to be used in production for the gui by the end of this week, as I understood it. is that what you understand, and is that still on schedule? Jon gave me an overview and it sounded like everything was well in hand with the possibly significant exception of mongo insanity. [16:35] That made me a bit nervous. [16:36] gary_poster, sinzui: specifically, as i understood from standup, we're still looking at the mongodb size which has given us problems. [16:38] Second topic was making sure that Orange and GUI are aligned in terms of incremental delivery. While it is true that we have a big deadline and unveiling for your work in a couple of months, the GUI has many incremental deliveries before then, and we want to make sure that Orange squad is comfortable and happy and aligned with the fact that their work will be released incrementally, frequently and much sooner than [16:38] a couple of months from now. [16:38] ack thanks for clarification jcsackett [16:40] gary_poster, we have sorted the db issue. We still see a performance issue relating to large data transfers that will be fixed, but I think there may be other undiscovered performance issues as hinted in the channel a few hours ago [16:40] sinzui, fantastic about db issue. [16:41] jcsackett: not sure if you saw but I talked jeff into letting my dirty tricks go, care to second please? https://codereview.appspot.com/9841044 [16:41] rick_h_: looking. nervous about "dirty tricks" though. ;-P [16:41] gary_poster, We also need to ensure all charm data is consistent. That is to say, charm rules have changed over time, but not all charms have been updates. I want this fixed this week. I think that will give us the reliability we need [16:41] jcsackett: bwuhahahahaha [16:43] sinzui, reliability fixed this week: also great. performance and reliability: will we have m.j.c updated our Friday morning so we can evaluate it and see if it is performant enough to reveal the charm browser (without flags) in the GUI? [16:44] yes, let's do that [16:45] cool thanks sinzui [16:48] sinzui, made appt for Friday. Gave you edit privs. Feel free to invite all of orange, move, or do what you will. You'll coordinate with IS to get m.j.c updated for then? [16:49] fab. Thanks gary_poster [16:49] ty [16:50] benji: I forgot to mention in the review that you need to add the new doc filename to docs/index.rst, for it to show up in the generated docs [16:51] teknico: that's a good one. I already submitted the branch though. I'll do a quick self-reviewed branch to add it to the index. [16:52] benji: sorry for the delay :-) [16:52] teknico: we really need a test that would have failed when I added a doc file but forgot to stitch it into the doc tree [16:53] and we need another test that shows that the docs build with no errors/warnings [16:53] benji: uhm, interesting notion. tests about docs? what next, docs about tests? OHWAIT [16:53] :) [16:54] jujugui: since having to recreate my juju-gui environment, make devel sometimes requires sudo...have i screwed something up, or is that just the way it is? [16:54] the former, I hope. I don't [16:54] that is new behavior to me [16:55] ok. i'll poke around and see what's happened. AFAICT it asks for sudo when doing npm get. [16:55] jcsackett: sec, got a fix for you [16:55] hurray [16:57] rick_h_: make this go away, and you will be my new hero. [17:00] jcsackett: sudo chown -R rharding:rharding ~/.npm ~/tmp/ [17:00] jcsackett: only s/rharding/yourUser [17:00] jcsackett: when you follow the hacking dock and sudo npm install the couple of deps, it creates those dirs owned by root:root and later it tries to reuse them for local use and can't [17:01] jcsackett: I hit it every time I setup a new lxc/etc [17:01] rick_h_: fixed it. fantastic, thanks. [17:06] adeuring: Then IMO "promulgated" should change to with the definition of "reviewed", because "promulgated" itself is not valuable info-- it is redundant with store_url. [17:14] * hatch lunching [17:14] abentley: Might make sense. The main issue is probably that we need to define better what promulgated/reviewed/approved/whatever means, and how we want to represent it in code and data. But I am meanwhile a bit tired, let's discuss this another day === deryck is now known as deryck[lunch] [18:06] benji: ping, I'm trying to lbox submit and the tests complete, but right after I get a error about tornado missing in test/websockietreplay.py? [18:06] benji: did that get added to a deps list or am I missing a manual step for lbox submit and the copy/dir it creates? [18:08] benji: https://pastebin.canonical.com/91804/ is the trace from submit. [18:09] benji: ah, nvm. It's a sys package so nothing would have installed it. any reason we can't use a download-cache/locked python version of these things? Isolate them more and not need manual steps to build? [18:09] benji: and curious how you got around the problem with the clone then so I can do this in an lxc vs my main system. [18:17] rick_h_: I'm back from lunch and readying the scrollback. [18:17] benji: rgr [18:17] ?? [18:17] benji: basically the how to get around the clone thing I think will fix things for now [18:19] rick_h_: I thought the recent PyPI CDN fix would have made that go away. Hmm, you may have cached broken downloads from PyPI. [18:19] also, my work-around was to install python-gflags into the system (apt-get install python-gflags, IIRC) [18:20] benji: yea, tried that but when I go to run a submit, it hits the virtualenv directory for it and fails to find it [18:20] we're in this strange with a venv, but with system packages mode that isn't sync'd right it seems [18:20] rick_h_: perhaps killing the virtualenv would help; I clobbered mine mightily while fighting the same problem [18:21] benji: yea, doing another make clean-all, but this is happening on lbox submit so it's creating the whole directory, merging trunk, etc all outside of my control [18:21] I would much rather use buildout for this. It is a bit harder to use that pip, but it has much nicer tools for making reproducable builds. [18:21] we use pip with it in charmworld. [18:22] http://bazaar.launchpad.net/~juju-jitsu/charmworld/trunk/view/head:/Makefile#L78 [18:22] ok, will keep hacknig at it then. I thought had you found a different fix. [18:22] I wish I could be of more help. [18:23] benji: so then if you don't have this what version of virtualenv do you use right now? I'm wondering if it's pre "all venvs will be --no-site-packages" change perhaps [18:23] benji: thus why your installing of the gfalgs system wide works while mine doesn't [18:23] err, gsflag or whatever it is [18:24] rick_h_: oh! that's probably it; on your advice I installed a newer virtualenv (in my account, not system-wide, but to the same effect) [18:24] benji: ok, so when lbox submit runs is it using the older system one or the newer local one? [18:24] newer, local === deryck[lunch] is now known as deryck [18:51] little concerned my laptop harddrive is cookin! [18:51] time for an upgrade! [18:51] I can feel the heat radiating off of it [18:51] cook cook cook [18:52] lol I have to completely disassemble the laptop to get to it. This thing needs to last until the new MBP comes out [18:53] * hatch wants haswell [19:02] yay, jenkins is happy again [19:05] i wish i could report i fixed it and this will never happen again. [19:06] bac: I accept your report and celebrate wildly :) [19:07] rick_h_: you better do it quickly before the next run of jenkins roulette [19:10] lol [19:19] :-/ [19:41] to add another script to the assets directory and have it combined in much like d3 do I simply need to add it to JSFILES in the makefile? [19:49] benji: on CI tests, i am seeing periodic failures on IE10 for set_test_result. the call to connection.request results in getaddrinfo failing trying to get info for saucelabs. (due to a bug it confusingly report "no such file or directory"). you have any thoughts? [19:54] hmm nothing i try will get the makefile to build angular into the all-yui.js file === BradCrittenden is now known as bac________ [20:00] O K what's the trick? [20:01] hatch angular? [20:01] to include assets into the bundle [20:01] hatch the js frsmework? [20:02] yeah [20:03] is the trick to create a link to it in the assets dir? [20:03] ok color me ingerested on whats up with pulling in another framework. [20:03] bah stupif d tablet keyboard [20:04] prototyping some stuff [20:04] but it's no good if I have to manually include a huge script tag in the index.html [20:04] lol === bac________ is now known as bac [20:27] jcsackett: ping, still around? [20:28] rick_h_: yup. [20:28] jcsackett: so I was poking at the network graph on staging http://uistage.jujucharms.com:8080/:flags:/browser_enabled/ and noticed that it's loading spin.min.js outside of the rest of the js? [20:28] jcsackett: is that expected or maybe a miss in adding the file to the list of known modules, merge-files, etc? [20:30] rick_h_: that's used for the intiial spinner when you open jujugui; i haven't monkied with that b/c my attempts to do so broke it locally. i have a cleanup branch that changes that initial spinner to use the spin YUI module and kill spin.min.js [20:30] jcsackett: ah, makes sense I guess. I don't have a before/after atm so it just jumped out at me [20:31] rick_h_: yeah, anything with spin.min.js predates my work--i left it alone while updating our spinners. [21:37] *dogwalk*