[03:13] huwshimi: hey in your proposal for the ie scrollbar you changed the height to auto [03:13] that won't work with the animation [03:16] hatch: Oh, let me take a look [03:16] if you want to do that you will have to use the max-height trick [03:19] hatch: Actually, I fixed the issue that required the height to be auto anyway [03:19] oh haha ok then! [03:19] hatch: The buttons were cut off in Firefox, but the other spacing stuff I did fixed that. [03:19] (I originally fixed it by setting the height to auto [03:19] ) [03:20] hatch: I'll push up the revert [03:21] cool, I'll review/qa once that lands [03:21] other than that how has your day been? [03:22] hatch: Going a lot better than yesterday :) [03:22] haha fixed the VM? [03:22] hatch: Yeah finally, but it look like it'll expire tonight which means setting it up again tomorrow. [03:23] hatch: Are there any more IE10 issues you'd like me to take a look at? [03:24] that's actually all of them that I know of [03:24] maybe do a once over to see if you can catch any? [03:24] hatch: Sure. [03:24] hatch: Change has landed [03:26] awesome thanks [12:28] jcsackett: did you figure out hte indent thing from yesterday? [12:42] morning bac and benji: when you have time, there is anothe guiserver branch ready to review: https://codereview.appspot.com/13153043 Could you please take a look? [12:42] frankban: i'll be glad to in a little while. [12:42] bac: thanks, no rush [12:49] * benji sips his coffee and looks at frankban's branch. [12:50] thank you [12:53] rick_h: yeah, i was using a different jshint binary for syntastic than our project. [12:53] so i just switched which jshint is on the path in my LXC. [12:56] jcsackett: ah, yea I'm doing the same thing but it was working so :/ [13:00] jujugui: can someone point me at a design doc that shows how the search bar should exist when the charmbrowser is completely minimized? [13:02] jcsackett: sec, looking [13:02] jcsackett: http://dl.dropboxusercontent.com/u/10004366/juju-gui/index.html is the idea from a demo I was looking at for animations [13:03] rick_h: huh. we have two separate cards for "search on canvas" and "keep search bar available", but this makes it look like that sort of needs to be done as one thing. [13:03] since there's no option for "search for charms in browser while minimized". [13:04] jcsackett: well, my understanding is the idea that the search box is always visible. And if you use it while minimized, you end up in /sidebar/search [13:04] e.g. it un-minimizes to show results [13:05] rick_h: except that animation shows "Search in canvas" when we minimize. [13:05] ...not even really sure why, though i guess that's vaguely useful when you have an environ with a lot of charms. [13:05] jcsackett: oh right, yea that's a second feature [13:05] to be able to search and load things from the environment [13:06] it kind of goes with versioned charms though so that we can load the data for the version deployed? I'm not sure if we want that incrementally or what [13:06] rick_h: but we do want to be able to search for things in the browser from minimized, and just load the sidebar when we have results? [13:07] luca...who's afk would know [13:11] rick_h: ah. well ok then. :-P [13:11] jcsackett: so I'd start with removing categories from the two viewmodes. then removing filters/updating the results output/charm containers, before going to the search deployed charms or search bar [13:12] as the 'how search works/is displayed' will be effected in all cases [13:12] rick_h: works for me. [13:13] rick_h: we're just removing the filter widget from those views, right? and any other bits of interacting code. filters themselves stay for AC. [13:13] so we're removing the filter UI, and we're not checking for approved results by default. Then we're changing how the search results are displayed. First reviewd in a container and then the rest in another container [13:13] jcsackett: sec, there's a good bug to look at [13:14] rick_h: ok. [13:15] jcsackett: https://bugs.launchpad.net/juju-gui/+bug/1202306 though I thought abentley had a post in there that had notes. [13:15] <_mup_> Bug #1202306: We need an "all" category [13:15] jcsackett: of course the end of that is waiting on UX. [13:16] jcsackett: but the work still stands to remove the filters, group the results in two sets/containers [13:16] rick_h: right, so i'll start a branch that kills the filter widget from the views. [13:16] jcsackett: k, but the display has to go along with it so that we can use the results. [13:17] because as soon as we remove the default filters the results go to crap [13:18] rick_h: i'm not sure i'm following. this bug conversation is all over the place, and i'm only sort of following how it links to what you and i are talking about. [13:18] * jcsackett may be having problems because he's not fully awake yet. [13:18] jcsackett: k, hangout? [13:18] rick_h: sure, one sec. [13:22] rick_h: hangouts seem to be not so good? [13:23] i'm disconnecting b/c i'm getting truly horrid static. [13:23] jcsackett: died on me, one try [13:23] jcsackett: one more try [13:24] luca: ping, do we have any UX updates on the search UI including the "all" category notes [13:33] jcsackett: oh, and did my test concern pan out? Or is it working peachy the way it is? [13:34] jcsackett: because that's more of a 'before this first branch lands' concern [13:34] rick_h: i put in the dones, and everything still passes. [13:34] so we're good. [13:34] jcsackett: ok cool. Yea they should pass, just wanted to make sure the asserts were getting hit [13:34] * jcsackett nods [13:34] thanks [13:34] yeah, that was a good catch. [13:35] I did that before where they passed, but never asserted :/ [13:46] rick_h: i suspect we have several tests like that. [14:23] jcsackett: so for the record, I just had to update my local jshint as well. Guessing there was some .jshintrc config differences between what I had installed vs the updated one from the repo [14:23] hatch: let me know when you get a sec. Want to talk this through. [14:23] sure umm lemme grab a coffee first [14:24] hatch: definitely, going to need your head unclogged :) [14:26] benji: i'm seeing failures in charmworld trunk. can you pull trunk and run 'make test' and see if you get them? [14:26] bac: sure, one sec [14:32] rick_h: guichat? [14:32] hatch: sure === rogpeppe1 is now known as rogpeppe [14:47] bac: do you have a few minutes to talk about the fix for https://bugs.launchpad.net/charmworld/+bug/1214627 [14:47] <_mup_> Bug #1214627: AttributeError when searching for "precise" [14:48] sinzui: yes [14:49] bac, guichat is available [15:14] bac: I forgot about my test run until now; everything passed. [15:14] benji: ok. i did a make clean, removing crusty old pyc files and the one problem test passed in isolation. running them all again [15:30] I see the order of the Featured charms have changed :) [15:35] jujugui reviews pleeeaase. https://codereview.appspot.com/12744049/ [15:35] Makyo: looking [15:36] rick_h, thanks. [15:43] Makyo: sure I'll take one [15:49] hey luca could you check out https://comingsoon.jujucharms.com/:flags:/serviceInspector to make sure the inspector passes UX? [15:50] jujugui call in 10 [15:50] kanban now. [15:50] hatch: sure, but won't have time to do it today, is it ok if I do it tomorrow morning (my time)? [15:51] you bet [15:51] do you have IE? [15:51] hatch: internet explorer? [15:51] yeah [15:51] hatch: I don't, do I need it? [15:52] guess not - we'll just make sure it looks the same in IE as it does in Chrome/FF [15:52] Makyo: got time after the call to hang on the call and chat about the MP? [15:52] hatch: ok [15:52] bac, .short_url -> .basket_name -> .basket.split() fails because .basket is None. There is a invalid charm in the system without a name, owner, or basket. [15:52] rick_h, sure. [15:52] luca: thanks, any issues plz email - I want to get it unflagged by the end of the week [15:53] sinzui: ahhhh. so can you delete that bundle? [15:53] and provide some defense? [15:54] bac. I can. I am preparing a query to see if the bad data is also in production [15:54] sinzui: i suspect it is [15:59] jujugui call in 1. [16:10] bac, The error cannot happen in production because the early ingest error was never in production. I will fix the class, prove it works on staging, then delete the bad bundle from the staging db [16:11] sinzui: great [16:22] hi adeuring, as i mentioned i'm seeing spurious failures unrelated (famous last words) to the work i'm doing. they are all tests you added recently in r357. could you have a look at http://paste.ubuntu.com/6010830/ [16:22] bac: sure [16:28] bac: where can I find your branch? [16:29] adeuring: lp:charmworld/official-bundle-json [16:29] adeuring: lp:~bac/charmworld/official-bundle-json [16:29] thanks [16:29] second is correct [16:31] * bac grabs lunch [16:31] adeuring: how long until your eod? [16:32] bac: officially, 30 minutes or so [16:32] adeuring: ok,ping me if you have any ideas. thanks. [16:38] Makyo: so how does your branch signify to the user that the charm can be upgraded? [16:38] I'm on rapi with serviceInspector and upgradeCharm [16:39] hatch, that's the next branch. This branch's whole point is to check for available upgrades. [16:40] gotcha [16:41] lgtm qa ok [16:52] jujugui CI is down [16:52] unable to deploy charm [16:52] were there any changes made to the charm yesterday? [16:54] hatch: https://code.launchpad.net/~juju-gui/charms/precise/juju-gui/trunk landed today it looks like [16:55] it had failed when huw landed his branch early this morning [16:55] ahh crap, gota run for a few [16:55] hatch: ok, well #93 landed yesterday [16:56] I'll look into it when i get back [16:56] #92 as well [16:56] BradCrittenden: do you know if we have any actual bundles in the production DB? [16:56] hatch: yes, but only the guiserver bits, and the error seems unrelated (KeyError: 'instance-state' when trying to parse juju status, wird) [16:56] weird even [16:57] benji: i do not. sinzui might as he just looked into it === BradCrittenden is now known as bac [16:57] benji, yes I do know. [16:58] sinzui: are there any bundles in the production db? [16:58] benji, bac, There is exactly one. Most likely the abentley won we are testing [16:58] darn [16:58] sinzui: that would've been my guess [16:58] benji, staging has 2, one is bogus and I will remove it after I have a fix in place for a bug [16:59] I guess I'll have to write a migration script then [16:59] benji, do you want to remove the production one? [16:59] it would be nice :) [17:01] benji, you could just remove() the one bundle, If you are updating the structure you could just make sure you are compatible during the period it is not reingested [17:02] hmm, yeah I think just removing all bundle data from ES and waiting for a re-ingest is the way to go; thanks [17:10] bac: I can't reproduce these failures :( [17:11] adeuring: thanks for looking. very odd. [17:11] adeuring: are you working within an LXC? [17:11] bac: no, do you? [17:12] adeuring: no. but i have seen spurious test failures in the past linked to using a "real" environment vs an lxc. those were timezone based, though. [17:12] bac: another odditiy: you started with trunk r358 -- and that revision already contains most of my recent work... [17:13] adeuring: well, i didn't run the full suite until recently [17:13] ah, ok [17:13] adeuring: anyway, have a good evening. thanks for your help. [17:14] bac: thanks. I'll look a bit more tomoroorw morning [17:14] adeuring: ok. i'll email you if i solve it [17:14] bac: cool, thanks [17:39] totally just cut my finger open on a PowerAde bottle.....awesome! [17:50] does anyone know the euca command to destroy an instance? [17:50] ^ jujugui [17:50] oh terminate-instances [17:53] ok another build kicked off [17:53] looked like it couldn't start up the new instance [17:53] will see what happens now [17:54] bcsaller: is there any way to QA your branch? [17:57] bac, benji: do either of you have time to review https://code.launchpad.net/~sinzui/charmworld/basket-name-none/+merge/181362 [17:58] sinzui: I can take it. [18:02] hatch: you can drag test/data/blog.yaml onto the canvas [18:03] ok and if no errors and the services are there then it's a o k ? [18:07] hatch: guichat when you get a sec please? [18:07] bcsaller: I think there were some things changed in your branch which will effect Makyo's re the charm promise [18:07] rick_h: omw [18:08] hatch: that is a wrapper around the normal call, that shouldn't be a problem. The id normalization could impact something else but shouldn't [18:12] hatch: yes, it should work, services and relations from the deployer file as you'd expect. [18:13] hmmm... though actually that won't work in the extracted version for the reason listed in the mp, ugh [18:13] when a deployer file includes more than one target we'd need a disambiguation UI which isn't there [18:14] so w/o a change, no, it won't import, it will throw the error saying more than one possible target [18:14] a deployer with one target will work though, an export for example [18:14] ok so....stop reviewing and qaing? [18:19] or just export something and re-import it [18:19] that didn't work before [18:19] when we unsynced the format [18:21] lol [18:21] * hatch is glad he isn't doing this [18:28] Makyo: if I wanted to confirm that the python environment .deploy() method would allow setting constraints in the 3rd argument (config) where would I look for the websocket docs/code? I see the send_rpc call, but want to keep following the code to determine what's valid for that config param for the 'deploy' call? [18:30] yay jenkins has been appeased! [18:33] rick_h, app/store/env/python.js:257 - Not seeing constraints :/ [18:34] Makyo: right, hatch thought perhaps the 'config' could container constraints [18:34] but the code doesn't read like it [18:34] Makyo: so I was going to go looking at the 'thing' that this ends up calling but not sure where to head from here. juju-core source? [18:35] jujugui I got CI back up, canonistack was holding on to an errored machine and wouldn't terminate it [18:35] Makyo: I don't see any tests that work like that either, but the tests are specific [18:35] cool [18:35] rick_h, Yeah, I'm not seeing it in fakebackend, either. It just uses an empty constraints object. [18:35] thats so classic canonistack, love that guy [18:35] Makyo: the goal is to deploy a charm from the ghost inspector with charm config + constraints. Seems like it *should* be possible but trying to find the right way to build that [18:35] rick_h, looking in core real quick. [18:36] thanks Makyo [18:36] bcsaller: lol [18:36] * rick_h is looking at core for the first time ever...duh duh duh [18:37] we have to be able to deploy a service with constraints :( That seems like such an oversight if it's missing [18:37] hatch: yea, in following the code there now it's not tested or something I am seeing. [18:43] rick_h, In core (haven't checked pyjuju), ServiceDeploy's params takes a Constraints argument: http://bazaar.launchpad.net/~go-bot/juju-core/trunk/view/head:/state/api/params/params.go#L58 [18:43] Makyo: ah, thank you sir! [18:43] rick_h, The constraints.Value shooould be a dict with very specific keys. [18:43] Will look. [18:44] Makyo: cool, yea I was looking at DeployServiceParams in conn.go which seems similiar, but not quite it [18:44] rick_h, http://bazaar.launchpad.net/~go-bot/juju-core/trunk/view/head:/constraints/constraints.go#L19 [18:45] rick_h, I hope this hasn't changed and I'm not leading you down the wrong path. [18:45] Makyo: ok, so looks like a little more work to do to get this able to work. [18:45] Makyo: cool, thanks. This helps a bunch and gives me something work towards. [18:45] rick_h, Yeah. Want me to check pyjuju too? Just got back from lunch, so I'm not on anything quite yet. [18:46] Makyo: sure, I want to peek as well to understand how this stuff fits [18:46] but appreciate a sanity check [18:48] rick_h, http://bazaar.launchpad.net/~hazmat/juju/rapi-rollup/view/head:/juju/rapi/cmd/deploy.py#L15 has a constraints_strs [18:49] I'm not quite sure how that works, but it does look like it's there. [18:49] Makyo: so for the config, the Go is caps, the python is not? [18:49] Makyo: yea, was looking at the constraints.py file [18:50] Makyo: actually guichat for a sec? Maybe easier? [18:50] rick_h, Correct; that should be taken care of in app/store/env/{python,go}.js, so the calls remain the same, as does the data passed to the callbacks. [18:52] rick_h, sure, be right there. [19:17] Makyo, its mem=4 cpu=2 instance-type=m1.medium [19:17] hazmat, cool, thanks. rick_h ^^^ [19:17] Makyo, w/ juju-core s/cpu/cpu-cores and instance-type doesn't exist [19:17] er.. mem=4G [19:25] sinzui/bac: can I get a review of https://code.launchpad.net/~benji/charmworld/bundle-heads/+merge/181385 ? [19:26] * sinzui looks [19:53] jujugui is there a way to get the env calls to follow their callbacks in tests? [19:54] the overview constraints page needs to send the new constraints, wait for it to return, then send it's new unit number [19:54] wrap it in a promise [19:54] but the callback is never being called in the tests [19:55] ehh, that's a lot of overhead just to make it pass the tests :) [19:55] and you're sure it not failing before that? or you think nothing is responding? [19:55] well the env constraints call is being made [19:55] but the callback isn't getting called [19:56] hatch: I can look at it with you but offhand I don't know what to tell you [19:56] guichat real quick? === BradCrittenden is now known as bac [20:03] benji, I replied with a question of sorts. [20:04] cool, looking [20:07] sinzui: I am inclined to say that bad bundles (or more correctly, bad baskets) are not ingested at all, that way we don't have to have a bunch of defensive code against bad data. How does that sound? [20:09] benji, how do we communicate to the basket author that their basket is invalid? [20:09] we have a report for charms. you cannot search for them, but the report shows what we know about them [20:10] benji, but in general I am fine to not save insane bundles for the moment. [20:10] * sinzui is writing an incantation to get the insane one out of staging ES right now [20:10] ah, good point. In that case I would save the basket data but none of the bundles. That way we can show a bad basket but none of the bundles will foul anything up. [20:11] we'll still have to be defensive against bad baskets, but that is a smaller battle front to defent [20:17] sinzui: how about that? ^^^ [20:18] benji, works for me. r=me. I'll let you work out the graceful exit from the situation [20:18] sinzui: "graceful exit"? [20:19] benji, adding the defensive guards for bad baskets [20:19] ah, gotcha [20:19] There won't be any in a few minutes I hope [20:19] bcsaller: is there an example in a current test about setting up the env with the real fakebackend? [20:19] I can't seem to find one [20:21] hatch: test_sandbox_python and look for integration, I think that does what you want [20:22] ahh got it thx [20:23] Then I think you pass that sort of a setup as env [20:36] oh jeeze I need to load up a fakestore too [20:44] bcsaller: ok so now that i Have it all set up and are monkeypatching onmessage I still only get a single call [20:44] I'm guessing onmessage needs to return something? [20:54] I really wish the testrunner didn't make me work to hard to craft the description of the subset of tests I want to run. I bet there is a plug-in somewhere that I'd like. [20:54] hatch: when you're env is using the sandbox connected to a fakebackend it should work [20:54] benji: nested describe blocks? [20:54] I use those to collect tests [20:55] bcsaller: got a second for a guichat? [20:55] bcsaller: this is the nose python test runner; it's not so bad, but it wants filename:classname.methodname. I'd prefer to just do methodname and let it figure out the rest [20:56] benji: ahh, different test runner, I've used tags for that with nose iirc, there is a plugin for that [20:56] hatch: missed you I guess [21:09] going to grab some lunch [21:42] sinzui: i've narrowed the problem to a small set of tests as seen at https://code.launchpad.net/~bac/charmworld/spurious-test-failures/+merge/181418 [21:42] * sinzui looks [21:42] i've listed abel as a reviewer so perhaps he can figure something out tomorrow [21:42] if not i'll land it so i can proceed with my work [21:43] * bac walks unusually patient dog [23:03] Morning [23:28] So this stuff is pretty good: https://twitter.com/bzr_pull_makyo/status/370264178293886977/photo/1 Nicely gingery without being too weird for a beer. [23:29] Plus, it's called Good Juju.