[00:24] hey huwshimi you have IE right? Can you QA something for me? [00:27] hatch: Sure [00:27] thanks! https://codereview.appspot.com/12683044/ [00:28] it fixes drag and drop in IE [00:28] I coudln't make it break but....ya know....ie [00:28] so I was hoping for another QA :) [00:31] hatch: Sure, give me a few minutes to get it all set up [00:31] yeah no rush, just comment on the review when you're done [00:31] Im not going to do anything more on it until tomorrow [00:33] ugh, the vm needs to do updates [00:35] hatch: Just QAing that we can drop a charm onto the "start adding charms!" message? [01:02] huwshimi: that's the specific bug that this fixes yes [01:02] IE has a odd issue where depending on where you drag from it'll either drag the whole element, or just one of the text blocks [01:03] that's not fixed and not a priority right now [01:05] hatch: Updates look like they might have killed my vm :( [01:05] Might have to rebuild it [01:06] Oh, now it's doing more updates [01:20] hey huwshimi, having fun? [01:21] rick_h: Soooo much fun [01:21] :) [01:23] haha that happened to me earlier this week [01:23] the updates just.....stopped it from working for a bit [01:33] hatch: Yeah working now [01:33] hatch: I can't get past the "Your browser is not fully supported" screen with the inspector flag turned on... [01:36] Or even with the flag turned off now... [01:41] Oh, I clicked some kind of compatibility mode thing [01:49] hatch: QAs OK [02:09] huwshimi: son of a B I did the EXACT same thing [02:09] that's a blog post! [02:10] seriously I spent like 20mins trying to figure out why the heck it wouldn't load [02:17] hatch: :) [08:19] gary_poster: pong [08:20] gary_poster: sorry about yesterday afternoon, had to leave a little early [08:20] teknico, np === schwuk_away is now known as schwuk [08:24] teknico, are you available for a hangout? [08:24] gary_poster: yep [08:24] teknico, ok will be using ipad so not guichat [08:25] gary_poster: do you want me to setup one? [08:27] teknico, I've tried to hangout with you twice :-P sure you try now [08:27] ok, switching google accounts... === schwuk is now known as schwuk_away === schwuk_away is now known as schwuk [11:04] yay friday [11:04] Hi jujugui. I am seeing this error for every test ftest when I test the charm on canonistack [11:04] ERROR Missing config 'username', 'password', 'project-name' required for userpass auth\n" [11:05] Any clues? [11:08] sinzui, OS_ env vars for creds [11:08] hmm. [11:08] * sinzui looks at novarc [11:09] sinzui, just source it, if you haven't already [11:09] env | grep OS_ [11:09] hazmat, I did. I think I soured the wrong one [11:28] hazmat, I have confirmed I did source properly. I think my env is not ignored my make ftest. I can see that running that will destroy the juju environment I previously bootstrapped and deployed to [12:46] luca__: ping, around? [12:54] gary_poster: luca__ I'm trying to match up the design image as best I acn without actual dimensions and the fact that the category icons don't look like the ones used in the mock up. http://uploads.mitechie.com/lp/autocomplete_design.png is where I'm at. Please give it a look over. I've got some tests to write and then I'll put it up. [13:42] jujugui review time please. Autocomplete for all. notes and qa instructions included along with link to the design reference. https://codereview.appspot.com/12545044 [13:42] rick_h: looking. [13:42] gary_poster: I'll hold off on landing until the screenshot is ok'd by design. [13:42] thanks jcsackett [13:45] rick_h: confused about the "is Category" check. is the id this is looking at a charm id, e.g. soemthing like "cs:foo/bar"? [13:45] jcsackett: sorry, previous branch. We manually add categories into the results and build a manual id to make them work with charm-tokens [13:45] maybe i'm not so much confused as surprised that "cat:foo/bar" is a valid charm id we get, if that's the case. [13:45] ah, ok. [13:45] jcsackett: that id is cat:~gui/ [13:46] so this is more about shoe-horning in categories so they work with the charmtoken for autocomplete display. [13:46] jcsackett: see line 158 in that file/diff (expand all) [13:46] jcsackett: correct, this way categories get an icon, behave/look like charm-tokens [13:46] rick_h: dig, that makes sense. thanks. [13:46] jcsackett: http://uploads.mitechie.com/lp/autocomplete_design.png for a screenshot (but please qa) [13:47] rick_h, here. just had one of those...interesting meetings. :-) jamie will look at it in a few minutes [13:47] gary_poster: hah, it is friday where the most interesting ones take place :) [13:47] ah, the ever growing use case of the charm token...who knew it would end up in so many palces ... [13:47] jcsackett: charm-token is the most reusable widget in the history of canonical widgets :) [13:47] rick_h: whoo! [13:47] (as long as you're willing to beat some data with a hammer for it) [13:48] jcsackett: hey, you have to agree to its API :) [13:48] fair. :-P === Guest____ is now known as jamieyoung [13:52] abentley: could you have a look here: https://code.launchpad.net/~adeuring/charmworld/1204985-address-key-error/+merge/179443 ? [13:53] adeuring: Sure. [13:58] rick_h: code looks good, QAing now. [14:01] rick_h: not a bug per se, but when i do 'apach' apache2-passenger is above apache2, which seems like an odd sorting. [14:02] jcsackett: yea, it's based on weighting from the server and such. We can file something upstream in charmworld but I just take it as I get it [14:05] rick_h: review up. LGTM, QA ok, some notes. [14:09] adeuring: Please update the comments to indicate that Charm.store_url has a revision and Charm.address does not. With that change, r=me. [14:09] abentley: ack, thanks [14:16] abentley: I'm doing final QA of my branch; how do I give myself admin access to my local charmworld? [14:17] benji: Admin access to the web site like charmers have? [14:17] yep [14:18] benji: log in. Anyone who's a member of juju-jitsu can log can act like a charmer on dev instances. [14:18] s/can log can act/can act/ [14:18] cool [14:20] benji: including staging. [14:26] abentley: "403 Forbidden"; from https://launchpad.net/~juju-jitsu: "You are an indirect member of this team" [14:27] benji: Wacky. I am also an indirect member of the team. [14:27] (but it works for me) [14:27] abentley: I'm visiting /charms/precise/apache2/featured/edit, if that makes any difference [14:28] benji: The login operation should work the same everywhere. [14:30] benji: I was able to mark a charm featured locally. [14:31] benji: I did it just now. (precise/apache2, in fact) [14:31] ok, thanks. I'm building a fresh trunk to see if it is a problem on my branch. [14:31] I get a 403 on trunk too. [14:32] benji: You get that error when you click "login" on your dev instance? [14:32] abentley: nope, after I login and I visit t/charms/precise/apache2/featured/edit [14:34] abentley: I figured it out: the team membership checkbox is not automatically checked when loggin in through two-factor auth so it wasn't sent to the app (that seems like something we should fix) [14:35] benji: We have it fixed for production. For dev instances, login.ubuntu.com does not trust your dev instance like it trusts launchpad, so it lets the user decide what info to communicate. [14:35] makes sense [14:35] in that case a big-ole message on login.ubuntu.com would be nice [14:37] benji: Sorry, I lied. We have it fixed for *staging*, but not *production*. [14:37] k [14:40] benji: What would the message say? [14:41] something along the lines of "UNLESS YOU CLICK ON THINGS BELOW, THIS PROBABLY WON'T DO WHAT YOU THING IT WILL DO" [14:44] benji: In some cases, the site will continue happily along, in other cases, it will prompt you for the missing data. Maybe we could trigger it on the teams extension, since that's a case where having the user supply the data later is not an acceptable alternative. [14:45] that would be nice [15:01] jujugui: weekly now, right? [15:02] jcsackett: I figured we were skipping weekly because of no Gary, and having the daily at the same time [15:02] benji: i'm just going by what's on the calendar. [15:03] pfft, calendars! the next thing you know you'll be wanting style guides and tests [15:03] 'by the book' suckers :P [15:04] jcsackett: sorry I was waiting until 9:#0 [15:04] keep it the same as the rest of the wk [15:04] because we wont' really have much of a retrospective [15:04] more time for coffee then [15:05] this has been a week of mutiny. [15:06] how dare you attempt to cage me with your schedules, limits, and rules! [15:07] lol [15:22] jujugui guichat in 8 kanban now [15:23] call at 9:#0 huh? I guess the capital "3" really emphasizes the time [15:24] exactly what I was going for! [15:29] jujugui guichat in 1 [15:31] jujugui starting without you [15:32] http://www.youtube.com/watch?v=4FfXLMpr_DY === marcoceppi is now known as marcoceppi|away [15:43] benji, bac: Do you want to have another sync call today? [15:43] abentley: sure, after lunch? [15:44] abentley: sure; how about after you review the branch that I haven't proposed yet? [15:44] bac: Works for me. [15:44] lol [15:44] bcsaller: https://codereview.appspot.com/12545044/ is the link for the review [15:44] bac: My lunch is in 15 minutes. [15:44] rick_h: thanks I was looking [16:30] bcsaller: can you do that reviewon the ie branch? https://codereview.appspot.com/12683044/ [16:33] abentley: when I visit /api/2/charms/interesting I get an error about "No mapping found for [downloads] in order to sort on". I expect there is something I have to run to create and populate that index, do you know what it is? [16:51] benji: charmworld/jobs/cstat.py will update that. We should really add downloads to the mapping, though. [16:51] abentley: cool, thanks [16:52] benji: np. [16:52] hey do we have a guide to get the Go dev env up and running somewhere? === abentley is now known as abentley-lunch [17:00] what the shit ie is now giving me an 'unspecified error' on identical code from yesterday [17:10] thanks bcsaller [17:11] hatch: you load IE in anger, you might be bitten :P [17:11] I'm so bloody pissed off right now [17:14] the error number is .... get this ... -2147467259 [17:14] who uses NEGATIVE Numbers for error numbers [17:18] abentley-lunch: I would appreciate a review of my feature breakout branch: https://code.launchpad.net/~benji/charmworld/featured-breakout/+merge/179503 === abentley-lunch is now known as abentley [17:19] benji: I just got back. Are you psychic? [17:21] I get that a lot. [17:25] benji: Your branch reintroduces the 'qa' attribute that I've removed. Is that an accident? [17:25] abentley: yep; ignore that I must have fixed some conflicts wrong [17:27] benji: at 215, I don't think CharmTestCase needs to derive from MongoTestBase [17:30] jujugui can someone with IE10 pull down my ie branch and QA it? I can't get this branch to run at all anymore, even on revno's that passed QA [17:37] benji: This looks good as far as it goes, but it doesn't introduce a migration, so it's not landable. [17:38] benji: Also, I think I suggested using a modified construct_charm_id to generate revisionless charm ids. I'm doing that in my upcoming branch, and it would be nice for the ids to line up. [17:40] benji, bac: I'm up for a sync-up whenever you are. [17:41] abentley: sure [17:41] abentley: guichat now? [17:41] bac: sure [17:47] I was finally able to get it working again... [17:56] abentley: I forgot to bzr add the migration file, I've pushed a fix. Lunching now. [18:46] abentley: does promulgated=true imply the owner is ~charmers? [18:46] bac: It is common, but not always the case. [18:47] abentley: i'm wedged then. in my current version i'm using the basket_id to be ~owner/basket/revno [18:48] abentley: for a promulgated bundle the url is /api/2/bundle/api/2/bundle/basket/revno/name [18:49] abentley: i mean: /api/2/bundle/basket/revno/name [18:49] Right. Well, I think it would make sense if the basket ID was not affected by promulgation. Would that help? [18:56] bac: An example of promulgated=True, owner !charmers is http://manage.jujucharms.com/charms/precise/juju-gui [18:57] abentley: right. so i need to extract the owner from the basket_id. [18:57] bac: I don't understand why. [18:58] bac: The owner is in the payload/basket_data, so why not use it directly? [18:59] abentley: for the promulgated case i showed above, the query is: [18:59] query = {'basket': basket_id, 'name': path[2], 'promulgated': True} [18:59] abentley: the basket in mongo will be prefixed by the owner name. in this lookup i don't know the owner's name [19:01] bac: I had hoped you wouldn't need to basket itself when retrieving a bundle. Can we fix it so you don't? [19:02] abentley: we have to support URLs like /api/2/bundle/basket/revno/name so we only have (basket, revno, name) to work with for the query. i'm open to suggestions [19:04] bac: Okay, so what data do you need, that's currently in the basket? [19:04] jujugui: is anyone else getting a string of 304s when they run test-server? i've just updated from trunk, and hadn't yet today. observing the behavior in trunk as well as my own branches that updated. [19:05] abentley: chat? [19:05] might be faster [19:05] abentley: I pushed my branch with the un-added migration file added. [19:05] bac: sure. [19:05] jcsackett: yup lots-o-304 [19:05] benji: I saw that, but not a fix for the QA stuff. [19:05] abentley: oh, I forgot that :) [19:05] fixing now [19:07] hatch: does it move past hitting those two files over and over with 304 for you? i just keep getting test timeouts with that. [19:08] GET /test/charm%20icon%20url 404 2ms [19:08] GET /juju-ui/assets/svgs/service_module.svg 304 5ms [19:08] lemme run the full suite [19:08] sec [19:08] jcsackett: the tests appear to be running as expected [19:08] hatch: hrm. not for me. [19:09] i do 304 on js-yaml over and over and then get a timeout. [19:10] benji: Also, I posted earlier: I think I suggested using a modified construct_charm_id to generate revisionless charm ids. I'm doing that in my upcoming branch, and it would be nice for the ids to line up. [19:10] yup 100% finished on a very slightly modified version of trunk [19:12] abentley: will do [19:12] benji: I am just putting up a branch that modifies construct_charm_id [19:14] abentley: this is how I did it: http://paste.ubuntu.com/5967438/ [19:14] benji: This is how I did it: https://pastebin.canonical.com/95719/ [19:16] abentley: given that we'll always be hardcoding the boolean parameter, two functions makes more sense to me; but it won't kill any kittens [19:18] alright, clean checkout time. [19:19] orangesquad, bac or benji: Could you please review https://code.launchpad.net/~abentley/charmworld/pre-versioned-ids-for-charms/+merge/179521 ? [19:19] abentley: I'll take a look [19:24] jujugui looking for 2 very quick reviews and a quick qa on https://codereview.appspot.com/12704043/ plz and thx [19:25] * hatch lunching [19:47] benji: I've never really heard this "always be hardcoding" rationale before. What's the virtue having two functions to avoid hardcoding? [19:49] the hardcoding rationale is mine, but the "a single boolean that changes the way a function works is an anti-pattern" thought has been around for a while [19:49] the virtue would be that having two named, straight-line (i.e., no loops or branches) functions is more straight-forward than one function that can do two things [19:50] this is a bikeshed, though, so we've probably already spilled enough digital ink over it ;) [20:24] bcsaller: replied, and class changes being re-proposed [20:24] hatch: thanks [20:28] bcsaller: ok repropose finished [20:30] jujugui also if there is anyone else available to review it, it's only like a 8line diff [20:31] hatch: I'll look [20:31] thanks [20:34] benji: lol [20:34] :) [20:35] thanks for the review - I'd like to use a real name but 80chars [20:35] I suppose I could store the string in a smaller var [20:38] hatch: you can use a longer name and split the line instead [20:40] yeah it's fixed :) [20:41] benji: do you know if we have a 'get started with go' guide for juju? [20:42] hatch: I don't think we have anything company-made [20:42] ahh ok - I was thinking this weekend I might give Go a go [20:42] I'll find something online [20:44] don't know what I'll make [20:45] what do people make with serverside languages now? [20:45] lol [20:46] everything I do has a gui [20:50] hatch: I prefer to think of them as non-JavaScript languages :-) [20:51] abentley: lol even my serverside is done in js now - but it's job is to respond to a gui lol [20:52] oo a console based todo app [20:52] that's what I'll do [20:53] jcsackett: you should probably make sure the tests CAN fail ;) [21:12] abentley, benji: you're both near EOD but if you'd like to do a review: https://code.launchpad.net/~bac/charmworld/bundles-display/+merge/179541 [21:13] bac: sure [21:13] thanks benji. i'll bbiab [21:22] bcsaller: so on the settings inspector page when I change a field it's background turns a light grey and I get a red star [21:22] after clicking save, it should clear those ui changes no? [21:24] hatch: yes, it should, the model should be set, that was working fine before [21:24] maybe I broke something moving the viewlets about, there was a bit of merging there for that [21:24] ok I'm on a fresh trunk checkout - the model is being updated on save [21:24] I'll look into in a few if you're not on it [21:25] I'm just working on the button bug right now [21:25] that might expose the issue as well [21:31] bcsaller: ahh this button issue does look related to the conflict stuff [21:37] dogwalk cut short by monsoon [21:42] bcsaller: hey there must have been a bad merge if this ever did clear on save [21:42] because it looks like that only happens on conflict resolution [21:43] would you have an idea of where that method may be? Or if it was just forgot [21:44] hatch: this is vs trunk or should I look at your branch? [21:44] bcsaller: might as well just look at trunk [21:45] I only changed one string so far heh [21:45] ok, checking into it now [21:48] hatch: it looks like the config/constraints callbacks do trigger clearChangedValues still [21:48] yep [21:48] but it's not responsible for clearing the UI [21:49] the conflict method in the viewlet is responsible for that [21:49] yeah.. its not bound [21:49] so do we have a slight architecture issue here? [21:49] which has already been invoked [21:49] looks like a minor one, I can cook up a fix for that [21:50] well I'm wondering if we shouldn't define a clear conflict UI method [21:50] so kind of normalize this conflict method a bit [21:50] I think in my testing I always pushed it down the conflict pipeline and that resovled it [21:51] oh ok, do you want to tackle this bug? [21:51] or would you like me to? [21:52] hatch: either way, tell me what you'd prefer [21:52] umm I can do it, lets have a quick chat in a minute to make sure I'm on the right page [21:53] hatch: thought syncedFields might be the proper place [21:53] look like you could loop the s and clear that state [21:53] in guichat [22:31] thanks benji === schwuk is now known as schwuk_away