[00:07] huwshimi: it's not in widgets. All templates are in juju.templates? /me has to check [00:07] var templates = Y.namespace('juju.views').Templates [00:07] huwshimi: ^^ [00:07] it's part of the "build all the templates at once" that kind of makes it have to be in views.Templates to work [00:09] rick_h_: Oh awesome. I copied and pasted but somehow got that wrong :) [03:41] jujugui, https://codereview.appspot.com/21440044 needs one review. Hopefully I won't be around to look at that review until tomorrow. :-) === gary_poster is now known as gary_poster|away [11:52] rick_h_: morning [12:06] bac: morning [12:21] hey rick_h_, nm, false alarm doing qa. the correct version of the deployer didn't install for me. [12:26] bac: oh, that's not good. Did deps not update? [12:26] rick_h_: i'm about to have to run to the vet. i'm proposing a change to charmworld to do logging when baskets are rejected. [12:26] rick_h_: i don't know. i removed the canary file and re-ran 'make deps' and then it was good [12:26] bac: ok cool. thanks [12:26] rick_h_: when broking local/lib/python.../juju-deployer only had a deployer.py file [12:26] none of the rest of the package [12:27] bac: hmm, ok. Yea that was the old 0.1.1 version I replaced in an earlier branch. [12:42] jujugui morning crowd, review request when the time opens up please https://codereview.appspot.com/21370044 [12:42] rick_h_: I'll take a look [12:43] benji: thanks === gary_poster|away is now known as gary_poster [12:58] good morning gary_poster; re. YAML vs. JSON: JSON is a subset of YAML, so unless our YAML parser is broken it should work [12:59] benji: that was my thought. I know the proof stuff 'just worked' that way. To inline the yaml it almost turns into json [12:59] benji, ah ok, good to know, thanks [12:59] benji, rick_h_ I was thinking we should review my notes and strategize in a few minutes? [12:59] +1 [13:00] gary_poster: sure thing, I've got the deployer code down and matching up email to source code atm [13:00] cool. will ping in about 5 or 10. thank you [13:15] benji, rick_h_ now ok? If so, https://plus.google.com/hangouts/_/calendar/Z2FyeS5wb3N0ZXJAY2Fub25pY2FsLmNvbQ.j0rk5d371ph8331ijtf48t2uj0 [13:32] benji, rick_h_ forgot one other suspicious thing to keep an eye out for: when I first did a ps auxwww | grep python, there were three instances of the same server. My guess/hope is that these were from the multiprocessing stuff, but another possible sharp corner to be aware of [13:32] hmm, ok [13:59] jujugui: question about patching. i need to do this: [13:59] with patch.object(job, 'proof') as proof: [14:00] but the method returns two things *and* i want to check calling args [14:00] the default object fails as it unpacks too few things and that form doesn't seem to allow a lambda [14:00] bac: you can provide your own double that will be called and can do whatever you want and return whatever you want [14:01] benji: so just forgo the loveliness of assert_called_with and hand-roll it? [14:01] bac: that's the only way I know to do what you want [14:02] ok, thanks [14:02] it is bad design, you have to opt out of the goodness to do something slightly unusual. A better design would grow with you. [14:10] benji: mind looking at http://paste.ubuntu.com/6364840/ and see if you can suggest a way to make it work? [14:11] what i have doesn't work b/c the scope isn't the same when the mock machinery actually calls it. [14:12] bac: first you have to port the entire codebase to Python 3 and then replace the "global" keyword with the "nonlocal" keyword [14:12] bac: if you don't have time to do that before lunch, you can make called_with a list and append the value to it to capture it [14:13] i bet that won't work. it still won't be in scope. /me tries [14:13] it will [14:13] trust the master [14:14] the problem isn't that it isn't in scope, it is that there is a new scope and your assignment is shadowing the outer scope's value instead of replacing it [14:14] you were right, i was wrong. [14:14] a blind man could tell you that [14:14] what the hell [14:15] benji: but when i break in the original there was no called_with [14:16] bac: I bet you had the "global" declaration in there, in which case it was grabbing called_with from the module namespace, but there wasn't one there [14:16] nope, i added the global later [14:17] hmm, I'm out of ideas then [14:24] benji: review? https://codereview.appspot.com/21880043/ [14:24] bac: sorry, pairing with Rick right now [14:25] well, that's bad as rick_h_ was my next choice! [14:29] lol [14:29] benji, rick_h_: this branch is blocking QA and release of charmworld. it is very short. if you can squeeze it in i'd really appreciate it. [14:30] Can I review reasonably? [14:30] bac: ok, we'll take a look [14:30] gary_poster: oh, i thought you were gone. you're all red in my client [14:30] that scared everyone away ;-) [14:30] sure, anyone can do it [14:30] ok cool [14:30] i'm not being pissy just anxious to move on [14:30] sure, that's what they all say [14:32] i was pissy when they told me at 9am the vet wouldn't be coming in until 11. that resulted in them agreeing to drive jojo home after the appointment. [14:32] bac: LGTM [14:32] ty [14:35] benji: when in rome [14:35] :) [14:52] hey rick_h_ , has your deployer file url branch landed? I don't see it in staging. If it is there, I will try to use it in the GUI [14:52] gary_poster: no, I was updating per review this mornign and haven't submitted it yet due to the burning barn [14:53] rick_h_, oh ok. mm. wasted work though. how much effort to get that in? can bac take it, maybe? [14:53] gary_poster: give me 3min to finish it up and hit rv-submit] [14:54] rick_h_, awesome, given. :-) thanks [15:01] gary_poster: submitting to the lander now. will take a min for it to run it through/merge it [15:02] cool, thank you. [15:09] gary_poster: ok, merged in successfully. Not sure how long it'll take staging to pick it up [15:09] awesome, thanks [15:09] gary_poster: so just using the deployer_file_url attribute in the api response should be good for that url [15:09] cool [15:09] on it [15:13] hey all [15:14] party [15:14] how's things? [15:14] on fire :) [15:14] good on fire or bad on fire? haha [15:18] when are our stand-up now that the time has changed [15:18] hatch: in 45min [15:18] oh ok cool . I didn't want to be Kate tomorrow if it was earlier for me or something [15:19] great picture from Robbie on the twitters :) [15:25] hatch: linky [15:25] * bac reboots charmworld staging due to memory issues [15:26] rick_h I'm on my phone at this conf so I can't but it was within the past 24h [15:27] hatch: cool, see it [15:29] rick_h there is a typescript talk this afternoon...thought you'd enjoy that :P [15:50] jujugui call in 10 [15:58] jujugui call in 2 [16:19] All throughout the call, the dogs play. As soon as the call stops, they crash. Go figure. http://ubuntuone.com/gallery/56atUQ5IHr3CkOXeT605zL/IMG_20131105_091512.jpg [16:28] Hey Makyo. Feel like taking a break to make a video? :-) We have a request from ODS [16:28] gary_poster, sure. What for? [16:29] Forwarding [16:32] Makyo, sent [16:33] Got it, thanks! [16:36] cool [16:36] thank you [16:48] gary_poster, do they want a version of jcastro's talk in video form, then? Drag the havana/swift bundle in simulator, while showing real deployments in EC2? [16:48] mark's talk you mean? he did that during the keynote [16:49] (that was very cool by the way) [16:50] Makyo, yes [16:52] * gary_poster is glad staging was not down at the time [16:52] that appears to be the issue now [16:57] oh I thought you guys planned for all that [16:57] so basically he did it off the cuff? [16:58] what did he do? /me hasn't managed to watch the video yet. He used comingsoon? [17:01] yeah he did it off the cuff. [17:01] rick_h_, will forward [17:01] gary_poster: cool, setting up in the background to listen to while deploys run [17:06] jujugui, small and easy: https://codereview.appspot.com/21980043 [17:07] (I hope :-P) [17:48] gary_poster, like https://docs.google.com/a/canonical.com/document/d/1mXcYaNHWP85o11UoiQZTXhgn5vt-5eU27fWgHFbIKGA/edit?usp=sharing ? [17:55] Makyo, sounds great! Thank you [17:55] Cool, on it, then. [17:59] rick_h_, you around? Would like to verify two apiv2 -> apiv3 changes. First, when searching, what was an api call with 'charms?text=foo' is now 'search?text=foo'. Second, what was 'http://localhost/api/2/charm/mysql-1/icon.svg' is now 'http://localhost/api/3/charm/mysql-1/file/icon.svg'. Does that look right? [18:00] gary_poster: not sure on the search one. Looking [18:00] thanks [18:03] gary_poster: that looks correct [18:03] awesome thanks rick_h_ [18:20] jujugui, I now have https://codereview.appspot.com/22000043 , which removes the charmworldv3 flag and APIv2 class. That's in addition to https://codereview.appspot.com/21980043/ , which still needs some reviewing love. I'm now going to try and get https://code.launchpad.net/~ya-bo-ng/juju-gui/linkify-charm-descriptions/+merge/192852tweaked and merged, and then I'm moving to changelog... [18:20] gary_poster: k, looking at the first one [18:20] thank you rick_h_ [18:24] gary_poster: should we decode jujucharms.com/bundle/%7Ejorge ? [18:24] the ~ [18:25] rick_h_, oh yeah I was going to and forgot, thanks [18:25] gary_poster: ok, LGTM on that one with that [18:25] gary_poster: also we need to make sure we switch off staging back to manage soon (don't want to have that fall through the floor) [18:26] rick_h_, +1. Once bac gets charmworld trunk deployed to manage then I think we can [18:31] gary_poster: jcsackett found a problem with charmworld that got turned on yesterday in staging. he's going to revert it. having that done is a blocker for updating production. [18:32] ack thanks for update bac [19:08] rick_h_, hey once I add those tests to that IE fix branch can it be landed? Did you QA? [19:11] hatch: no [19:11] oh ok, think you'll have time for that today? or are you swamped? [19:12] on fire sorry [19:12] alrighty np I wont rush to finish the tests today then :) [19:12] * hatch hands you a bucket of water [19:12] just in case you get too hot :) [19:18] jujugui: review por favor: https://codereview.appspot.com/21790045 [19:19] bac, would do it but I think one of the charmworld people would be better for those topics [19:20] abentley: i just wrote my first migration for charmworld. would you mind taking a look? https://codereview.appspot.com/21790045 [19:36] bac: Sorry, I'm gonna be a while. [19:36] abentley: ok. let me see if sinzui is available [19:37] bac: No, we're working on some stuff right now. [19:37] oh [20:24] gary_poster: ping for 'wtf is going on' catchup? https://plus.google.com/hangouts/_/calendar/Z2FyeS5wb3N0ZXJAY2Fub25pY2FsLmNvbQ.j0rk5d371ph8331ijtf48t2uj0 [20:48] bac: I think when es-updates creates a new index, it uses an old index, not mongodb, as its source. And it does this only when there is an incompatible schema change. [20:49] abentley: so you're saying the old data will remain in es? [20:49] bac: Yes. [20:49] abentley: advice on the right way to proceed? [20:51] bac: I think drop all the bundle entries from elasticsearch. [20:52] bac: The bundles are going to disappear until the next ingest and that's okay, right? [20:52] yes, that's what we want [20:53] bac: So yeah, I think that's the thing to do. But do you need to delete the bundles at all? [20:55] abentley: many of the bundles that are in charmworld have errors so we'd like for them to be removed so that after the next ingest only error-free bundles are presented. does that answer your question? [20:57] bac: Just raises more questions, I'm afraid. The policy with charms is that proof failures do not prevent them from being stored and displayed. [20:58] bac: The store was what determined whether a charm was acceptable to display. [21:00] bac: We did at one point hide charms with proof errors, but stopped doing that because proof errors did not mean that the charm was unusable. [21:01] bac: I believe marco is changing it so that errors indicate that charms are unusable. [21:01] abentley: The store does not know about bundles. It is our decision for the first release of bundles to reject those with errors. At a later date, when we've had time to update the UI, we will store those in error and show the problems. [21:04] bac: If it were me, I'd hide those with errors, rather than preventing them from entering the store. What happens when proof changes and some of the bundles currently in the store start to fail proof? [21:06] abentley: that's a good point. [21:09] bac: Anyhow, that's my 2 ¢. You do need to delete bundles from elasticsearch, but changing the way you handle errored bundles can wait. [21:10] abentley: thanks, i really appreciate your input. [21:10] bac: np [21:28] benji, hey. I exported from comingsoon and got this: http://pastebin.ubuntu.com/6366975/ [21:28] * benji looks [21:28] benji, it looks like it is right. I wonder if this is because jujucharms.com is old [21:28] ooh, that's good [21:28] or if people are exporting from pyjuju or something [21:28] might be, at least it means we are closer to a global fix than we though [21:28] right [21:29] so, you agree that we appear to already be in good shape here, right? [21:29] in terms of our release goals [21:45] gary_poster: replied to your email, but correct. So we're just on the charm and deployer bits to move forward [21:46] cool thanks Rick. I'm writing a status email for the peeps list [21:59] Morning [21:59] hi huwshimi [22:00] hey huwshimi [22:01] benji, re your card "Not-quite-private-enough charms were loaded into charmworld. People want them out.": that is in urgent already [22:01] deleting [22:01] gary_poster: k [22:01] I searched for a card with that issue number and couldn't find one, so I made that one [22:01] gary_poster: aaron suggested some changes so could you do a final of https://codereview.appspot.com/21790045 [22:02] bac on it [22:13] bac, sorry, oven fire and other excitement. back and then must run. trying to finish... [22:14] bac LGTM [22:15] * gary_poster running === gary_poster is now known as gary_poster|away [22:19] rick_h_: Would it be a bad idea to have custom templates per instance of the dropdown widget? As in do something like TEMPLATE: Y.namespace('juju.views').Templates[this.get('template')]; (which doesn't work as is... but you get the idea) [22:21] Another option would be for the widget to not handle the template at all, but just handle the show/hide [22:23] I guess that might work well with the need for rather different content for each dropdown... === gary_poster|away is now known as gary_poster === gary_poster is now known as gary_poster|away