[01:31] gary_poster, done [12:02] gary_poster, hi there! [12:05] hey matsubara :-) [12:05] thanks hazmat :-) [12:06] gary_poster, do you have a MP that I can test the tarmac setup? [12:06] mm [12:06] looking [12:06] teknico, hi. you around, and do you have an MP for the CI test failures? [12:07] matsubara, to add to the fun, some tests were failing last night :-P :-) [12:07] gary_poster: I sent you what I had via email [12:07] teknico, ok looking [12:09] gary_poster, yeah, I saw that. that's why I waited to check with you first. once there's a MP available I can run tarmac manually to check it will do the right thing this time [12:10] matsubara, do you want a stub MP that can't land because tests will fail? [12:12] gary_poster, could be. if you're sure the tests will fail, then it should be ok [12:12] teknico, before I start investigating myself, was there a reason you chose to delete the tests other than expedience? [12:12] matsubara, ok one sec. I'll make a doc fix of some sort and put it up. [12:13] thansk gary_poster [12:13] gary_poster: no, just that I did not know right away how to fix them to use the charm browser instead [12:13] teknico, did you look at Jeff's branch? [12:14] gary_poster: I did [12:14] were there problems in that approach, or do you think I should pursue it? [12:15] gary_poster: what approach? that branch is merged, it removes the charm panel for deployment [12:15] gary_poster: I'm missing something here [12:17] teknico, no, Jeff had changes in a newer version of that branch that looked like they should have fixed the problem. Go to branch and click on most recent revision https://code.launchpad.net/~hatch/juju-gui/remove-charmsearch [12:18] gary_poster: oh wow, that was unexpected :-) [12:19] ok, teknico. I can see how my email was not clear enough. sorry about that. I'll take over from here. [12:21] gary_poster: yes, the changes in that revision seem reasonable [12:21] cool [12:21] gary_poster: do you have time for a quick hangout? [12:22] teknico, in 20? [12:22] gary_poster: sure, thanks [12:43] teknico, fighting lbox etc. almost, will ping [12:43] gary_poster: ok [12:44] gary_poster: teknico let me know if there's anything we can do to help as well. Haven't peeked at it yet, but maybe help find a good replacement for whatever check was made [12:45] thanks rick_h. if it's fast for you, you could take a look at http://bazaar.launchpad.net/~hatch/juju-gui/remove-charmsearch/revision/728 and see if it looks right to you [12:46] gary_poster: ok, so the plan is load the sidebar, click on a charm, click on the add button and make sure things render at each step? [12:47] rick_h, yeah.\ [12:47] rick_h, note that this is supposed to be a utility [12:47] that deploys a charm [12:47] gary_poster: k [12:47] we do this for a number of tests [12:47] https://code.launchpad.net/~gary/juju-gui/restanalytics/+merge/169407 [12:47] matsubara, ^^ [12:48] matsubara, will be fine if it lands :-) [12:48] teknico, guichat? [12:49] ugh I need to restart soon. chrome is unhappy [12:49] gary_poster: yep [12:53] gary_poster: so yea, the approach looks good. I'd make sure to add notes in this that it's all depending on manage.jujucharms.com and any issues getting to that will cause the charms to not load. I'd also probably put notes in the specific templates that these values are used for CI and changes must include syncing the CI deploy method. [12:53] rick_h, good idea. thank you for the review! [13:06] gary_poster, sorry, was on a call [13:06] np matsubara [13:23] teknico, hatch fwiw I will be working with a variant of Jeff's branch that has some changes per Rick. lp:~gary/juju-gui/fix-charm-panel-tests [13:28] gary_poster: ack [13:44] gary_poster, did you see the error in the MP? [13:45] matsubara, looking [13:45] matsubara, that's canonistack insanity [13:45] matsubara, look at bottom of https://jenkins.qa.ubuntu.com/job/jujugui-merger-trunk/11/console [13:46] matsubara, "RETRY THIS TEST RUN!" [13:46] the next one failed too [13:46] https://jenkins.qa.ubuntu.com/job/jujugui-merger-trunk/12/console [13:47] matsubara, another canonistack insanity :-( [13:47] I'm retrying anyway [13:47] matsubara, tarmac will force us to switch to ec2 or hp cloud :-) [13:53] gary_poster, looks like it's going to work this time [13:53] http://10.189.74.2:8080/view/CE/job/jujugui-merger-trunk/13/console [13:56] matsubara, yay! matsubara what is diff between http://10.189.74.2:8080 and https://jenkins.qa.ubuntu.com/ ? I can log into first, but not second [13:57] internal instance and the public instance. they have different log ins (the public one you don't need to log in) [13:58] and it's just the published result, all the job config is in the internal one [14:04] ah cool [14:05] morning [14:05] gary_poster: ahh so that error was real, it can't get to manage.jujucharms ? [14:07] teknico, hatch on call but Jeff's work seems to almost work. watching it on browser, the only problem is seeing that thing has deployed [14:07] on call will contonue soon [14:09] matsubara, that error is good! [14:11] matsubara, the only thing is that we don't have proof that the right branch was tested...oh interesting. We set origin to my branch [14:11] matsubara, that's the only problem I think [14:12] yes, that's what I fixed. now the tarmac jenkins plugin sets the juju_branch before the test [14:12] matsubara, because that means we are not testing merge though? [14:13] rick_h: updated the charmworld icons MP. care to take another look? https://code.launchpad.net/~jcsackett/charmworld/reviewed-icons/+merge/169031 [14:14] what do you mean? tarmac merges your branch into lp:juju-gui, uses that to start bin/test-charm with your branch as the origin [14:14] jcsackett: sure thing [14:20] gary_poster: the error when deploying the charm with juju-core and the options set for CI tests was about staging, not sandbox, sorry: http://pastebin.ubuntu.com/5764804/ [14:25] abentley: got a sec to chat? [14:25] rick_h: sure. [14:35] is anyone else seeing test failures on trunk; specifically a timeout in 'application hotkeys "before all" hook' [14:35] I have success locally on unit tests (still on call) [14:36] benji: I've seen that to sometimes [14:36] *too [14:36] woah a new system 76 laptop that isn't massive [14:37] little on the heavy side though for an 'ultrabook' [14:38] buuuut you can get it with 16GB of ram [14:38] hmmm [14:41] jcsackett: feedback on the way [14:43] hatch: 1080p on a 14"? wow. if only it was from somebody else... ;-) [14:43] lol, not a system76 fan? [14:43] I don't really know anything about them [14:44] <3 my system76 desktop :) but I'd not buy their laptops personally. [14:44] I've never owned one but at a former place of employment we had endless problems with a half dozen or so of them. [14:44] great guys, did an interview with their CEO at PyOhio a few years ago [14:44] it looks a little plasticy [14:45] yea, the laptops just aren't the things of legend. [14:45] the price is right though... [14:45] I do <3 that they're colohugging all the displays that they ship though [14:45] great move, wish all laptops did that [14:46] I "need" a new laptop heh so I'm definitely looking at everything :) [14:46] colohugging? [14:47] https://plus.google.com/104919222657565747428/posts/7TxBM96MZYX [14:47] http://www.hughski.com/ [14:47] have one and love it for my various displays. [14:47] ok sorry back [14:47] ahhh cool [14:47] my monitor was calibrated out of the factory like that too [14:48] matsubara, you said "tarmac merges your branch into lp:juju-gui, uses that to start bin/test-charm with your branch as the origin" [14:48] my concern is that the behavior is this: [14:48] tarmac merges brancg into juju-gui *locally* [14:49] then tells the test-charm sccript to run tests on *original* branch (not merged) [14:49] In order to do the right thing, AFAIK, given the current code, we would need to do something like this: [14:49] tarmac merges branch into juju-gui locally [14:50] *tarmac pushes merged branch with --force to standard branch that it controls* [14:50] tarmac uses --origin=*that standard branch that has the merge* [14:51] Because tarmac is effectively single threaded [14:51] that is safe [14:51] jujugui call in 9 kanban now [14:51] thanks Makyo [14:52] teknico, I think there's some other way to start these tests that makes the problem go away [14:52] I'm not sure what it is [14:53] but if you could send a note to juju-gui with the problem I bet someone, possibly frankban, will give the answer between now and when you return :-) [14:53] gary_poster: right, will do [14:53] thx [14:54] hatch did you catch what I said about what you had done? it all works except for the part you didn't change, afaict :-) maybe can pair on it briefly; I think we are almost there [14:55] hatch: IIRC, both frankban and Makyo have system76 notebooks and neither is very pleased with them [14:57] I had one and returned it. I thought frankban was reasonably happy except size [14:58] Mine had a bad stick of ram and a color calibration issue fixed in 13.04. Other than that, yeah, it's big, and the battery isn't great, but I like it well enough. [14:58] gary_poster: nope, it overheats and turns off under not very heavy usage [14:58] jujugui call in 2 [14:58] teknico, oh :-( [14:58] Not sure I'd get another, though. [14:59] gary_poster: haha that's...good? So the auth failures were? [14:59] no idea hatch. did not dupe here [15:04] luca___, hi. where is the user data form effort? [15:05] gary_poster: Peter is currently working on it, I'll check in with him soon. [15:06] thank luca___ [15:06] gary_poster: going to send you wireframes in a few mins for a proof before they go to the team [15:06] yay! [15:16] gary_poster, I see what you mean [15:26] gary_poster: https://docs.google.com/forms/d/1Me4macMBX6KydCYyCC6YEjKnpCKakQCYlGlGxIB6Z0w/viewform [16:07] luca___, still on call. form looks like it has a lot of good stuff. Can we have edit power on it? I need to fix a few things and also add the legal language at the bottom [16:38] sinzui: displaying icons in a non-brittle fashion on charmworld is starting to look like it may consume more time than expected; is it worth continuing, or should we drop it in favor of other work? [16:48] there is 3" of rain in my wheelbarrel since yesterday and the forecast is for rain until Monday [16:51] so, I'm still not clear on the problem, much less on a solution, and I'm about to EOW (and EONW too :-) ), so I need to delegate this loadFixture thing to someone. rick_h, hatch, any takers? some details in the card in Maintenance [16:54] teknico: I should probably stick with the story I'm working on for now [16:54] teknico: yea, can't move off atm and the problem is temporarily worked around. I can chat with you on Monday if you want as I think there's an easy but semi-fragile fix we can try. [16:55] teknico: but toss it out to see if it's 'better' than what we have now [16:55] rick_h: the thing is, I'm off next week :-) that's why I' trying to get it reassigned [16:55] teknico: ah, well in that case... [16:59] grr, the 50 character lbox commit message limit is irritating [17:00] ok, it's email then [17:01] benji: you said it- it woudln't be so bad if it had a counter :) [17:01] hatch: use a real ide :P [17:02] hatch: well editor, vim position display ftw [17:02] s/ide/editor/ ;P [17:02] yea, got myself. I've got a bad case of friday's today letting things slip [17:02] I like my development environments disintegrated, thank you. [17:03] lol [17:03] I have been looking at webstorm for a long time [17:03] can't really see it would gain me anything over sublime though [17:11] hatch auth tests pass fine on saucelabs for me on IE 10 so far... [17:13] hmph intersting, it failed multiple times for me [17:16] If they pass then that's all that matters :) [17:18] hatch, others: reviews pls https://codereview.appspot.com/10253055 :-) [17:18] rick_h, maybe? ^ it is what you looked at this morning, with changes from you and to get it working [17:19] gary_poster: looking [17:19] thanks [17:20] I figured out my timeout issue. It was a combination of my own forgetfulness and the crazy world of web development: I had arbitrarily slowed down my network to better see the results of my branch, but of course the test runner does a bunch of HTTP, so it timed out. [17:20] heh [17:20] glad its resolved [17:20] it's [17:21] I'm taking Nicola's card. [17:22] jcsackett: http://paste.mitechie.com/show/966/ for the record. Manual url building ftw I guess :( [17:22] I'm also wondering why his email went to my personal account. [17:22] jcsackett: pushed to bzr push lp:~rharding/charmworld/reviewed-icons [17:24] benji, I checked the mailing list and you are subscribed to your canonical account. I am disinclined to suspect mailman, so I wonder if it is something google's side? [17:24] hmm, that's a possibility [17:26] this is an odd bit from the headers of that message: "Delivered-To: benji.york@gapps.canonical.com" [17:26] gary_poster: LGTM + notes on the way [17:26] cool thanks rick_h [17:31] jcsackett, how is it brittle? I thought the cases were: A is_approved and has icon,, use it. B. Fall back to icon for category. C. Use misc if category is not one we know. [17:31] sinzui: I've got a fix going up that cheats and should work lp:~rharding/charmworld/reviewed-icons [17:32] tests pass and should be able to just go from there. [17:32] fab [17:38] gary_poster: on it, sorry I missed the ping [17:51] sinzui: brittleness was in generating the image url. [17:51] i was building manually, which will break on the next API upgrade, but the routing stuff was mangling things. rick_h solved the mangling. [17:52] oh, understood. Thank you jcsackett [18:00] * hatch lunching [18:05] sinzui, have you all done any load testing on charmworld? [18:06] gary_poster, not on production. [18:06] gary_poster, On staging, I crawled it to find all the 500 and 404 and saw it never failed. [18:06] gary_poster, we started discussing how to load test charmworld this week [18:07] sinzui, fantastic. would love to see load test, focused on interesting and search results. I'm occasionally seeing missing editorial errors on gui, but not wure what it means [18:07] oh, yes. I see that two [18:07] gary_poster: is that true as of this week? [18:08] rick_h, yes, saw today [18:08] gary_poster: we made a change this week to help with that on editorial at least. [18:08] gary_poster: k [18:08] gary_poster, and when I reload it is there. There is a 3 second timeout between squid and the app/db/search. [18:08] hm [18:08] well, it sounds like you guys are on the trail of it. thank you. [18:09] we want to make search faster, introduce etags on the results. [18:10] cool. reliability of the sort we are talking about might be higher priority now, though. search speed is not a problem for me IME, though faster is nice [18:11] understood [18:11] gary_poster: if you get it again. Please check the network tab in the browser and look at the return code and the time of the request. We were thinking of chasing some middle layer issue in the squid/nginx/etc that seemed to pop at the 3s interval and I'm curious if you're seeing that or not. [18:12] interesting should really never take 3s any more though :/ [18:12] ack rick_h will do. last time was in the middle of automated tests so I didn't get a chance to analyze [18:12] rick_h, there is no nginx...and I think its smart caching was helping us. [18:12] sinzui: sorry, apache. You're right. [18:13] gary_poster: understand. appreciate keeping an eye out for it. [18:14] I had to configure apache to force http 1.0 because pyramid is vulnerable to long request. The other option was to switch to gevent [19:13] matsubara, hey. so, more tarmac next week? I think CI tests should be working again [19:13] should I re-enable the old jenkins project [19:13] so we at least have post-commit style CI for now? [19:14] gary_poster, yes, I'll re-enable. just a sec [19:14] thank you [19:15] gary_poster, done. [19:15] thanks matsubara :-) have a great weekend [19:15] thanks you too! [19:33] sinzui: mid-imp? [19:33] abentley, I can in 30 minutes. I am just leaving to get children [19:34] sinzui: okay. [19:46] yay jenkins is back to normal [20:11] Hi abentley, I am ready to chat [20:30] rick_h, jcsackett, are one of you testing staging and icons? [20:30] Do either of you want to see the error? [20:30] sinzui: i am, and i would love to. [20:31] jcsackett, http://pastebin.ubuntu.com/5765889/ [20:32] jcsackett, That is the only error. I have 3 occurrences in my inbox. [20:32] sinzui: yeah, and i just accidentally triggered another when meaning to open the same path on manage instead of staging. [20:33] oddly, going to the generated url locally or on manage causes no problems. [20:34] orangesquad, jujugui, we have +1 from Robbie on sprint for North American team members in Raleigh, NC July 15-19. I am filling out the paperwork to make official, so it is not time to buy tickets, but it is time to check dates and let me know if there are any issues. [20:34] gary_poster: i' [20:34] gary_poster: i'm just not sure i'll be able to get to that location. [20:34] heh [20:34] jcsackett, :-P [20:35] I'm just glad it will be a short flight. [20:36] gary_poster: LGTM if I can bring my kiteboarding gear and take a day off if it's nice to goto Jordan or Falls ;) [20:36] hatch: i don't think kiteboarding is allowed at Jordan Lake. [20:36] no? is it too awesome for it? [20:37] the lake can't handle that much awesome in one spot? [20:37] heh, maybe so. There are a couple of decent sized lakes within 30-45 minutes. Worth exploring at least. [20:37] Weekend would be good ;-) [20:37] gary_poster: I have stuff on Saturday the 13 and Saturday the 20, but during the week, all I have scheduled is taking out the recycling. [20:38] heh ok :-) [20:38] rick_h: there is a Lake Michie in NC....coincience? [20:39] thank you gary_poster. oh. and I will be 30 minutes closer to tRaleigh, this time [20:39] :-) cool [20:39] the sprawl around Raleigh is pretty crazy [20:39] must be lots of 'bedroom communities' [20:41] sinzui: i cannot figure out what combination of things has caused this. shall i rollback for now? [20:43] let me look before recommending rollback [20:43] jcsackett: do you live in raleigh as well? [20:43] hatch: i live in durham, which is another city in the same metropolitan area. [20:43] ahh one of those bedroom communities I was talking about [20:43] hatch, that's part of the sprawl. it's not really bedroom communities. they have those too [20:44] hatch: i'm not sure what you mean by bedroom communities. if you mean like "suburbs", not so much. [20:44] but Raleigh Durham Chapel Hill are full fledged places which come together into "the Triangle" [20:44] oh - here a bedroom community is another city...say within a 30min drive of the city where people live but don't work [20:44] I mean there are some places to work usually but not many [20:44] Bedroom communities and/or suburbs would be Cary, Wake Forest, Apex, and so on [20:45] no kind of more confused than that here :-) [20:45] hatch: ah, i see. still not quite accurate; each one has quite a lot of live/work. :-) [20:45] ahh gotcha - well here there is 1M people in the whole prov so that's to be expected I suppose :) [20:50] holy smokes that's a long ways from me [20:50] 8+h flight times haha [20:52] and when you get here everyone'll probably melt [20:52] it is *very* warm. [20:52] though today has been nice, post-storm. [20:52] well I was in NOLA for a while and survived ;) [20:52] Yeah, spring or fall would show off the area to better advantage [20:53] you are sort of picking the worst season for people to come, given the group has canadians. :-P [20:53] Canadian....single :P [20:53] or is someone else one too? [20:53] hatch: abentley is also from canada. [20:53] Aaron [20:53] ohhh right I forgot he is on our team now [20:53] well [20:53] he is orange [20:53] and orange is helping us for as long as I can keep them [20:53] and rick is very, very close to y'all's border. [20:54] abentley: is from the east though [20:54] ...the east... [20:54] they think Canada ends at the Quebec border :P [20:58] jcsackett, I think you should rollback. There is some interesting rearrangement of data happening between you route additions and views.api.API2.{__call__, _handle}. I don't think the path is being unpacked. [21:06] jcsackett, I am going to put a pdb in .charm() and make the same request to see what was really passed to the method [21:06] hatch: My mom's from Edmonton and my sister lives in Vancouver. It's not like we're disconnected from the rest of Canada. [21:10] abentley: so do your friends think that your family lives in another country? [21:10] :P [21:11] hatch: No, at least half of my friends weren't born here either. [21:11] haha [21:12] a couple guys/girls i know moved to TO and they said everyone they ran into thought we were all farmers [21:14] sinzui: yeah, i think the route stuff is wrong; need to talk with rick_h about that on monday. [21:15] jcsackett: after deploying a charm the url shows /sidebar/ even though / is the default to show the sidebar and /minimized/ means it's hidden is this a bug? [21:16] hatch: on uistage? [21:16] i see '/'. '/sidebar/' is still a valid url, just not the default. [21:17] right, click a charm, click 'add' the url is now /sidebar/ [21:17] should it not be / [21:17] ? [21:18] hatch: well, it's not invalid. what it should be i don't actually know. [21:18] but yeah, probably just '/' -- file a bug for it, would you? [21:18] * Makyo dogwalkify. [21:18] sure thing - I wanted to just check with someone who might know before I did :) [21:18] jcsackett: sinzui ah, I see the error. Please rollback [21:18] hatch, jcsackett it actually needs to be configurable IIRC [21:19] rick_h: in the process. [21:19] when juju gui is used as tool, sidebar should be root [21:19] jcsackett: sinzui the tests passed but yea, most tests hit the direct method call :/ I'm sorry. Did a very basic QA [21:19] rick_h: i swear it worked locally, but i must have been looking at my branch pre-merge. [21:19] when used as jujucharms.com it should be fullscreen [21:19] jcsackett: sinzui I thought the route diff was ok, but now see that I was mistaken. Sorry for the trouble :( [21:19] rick_h, is it config.add_route('api_2', '/api/2/{endpoint}/*remainder') [21:19] at least that's what I think we want/said [21:20] rick_h: all good. we can rollback now, and make the fix monday morning. [21:20] sinzui: yea, I did that to make the route_path method work correctly. [21:20] sinzui: because it can't handle a route like the original [21:20] jcsackett: thanks, apologize again for the trouble at the end of a friday. [21:20] rick_h: as long as it happens before I EOD, all good. :-P [21:21] sinzui: MP up for you to glance over. [21:21] heh, well I just finished mowing the lawn so ugh to walk into I broke it [21:21] sinzui: https://code.launchpad.net/~jcsackett/charmworld/rollback-266/+merge/169546 [21:22] r=me, we can wait for tarmac [21:22] whoo. [21:23] gary_poster: ahh so....this isn't really a bug per-se it's just a not fully developed feature? :) [21:23] jcsackett: I bet it's because *reminder returns a list while remainder:.* is a single entity. [21:23] jcsackett: a string [21:23] hatch, eh file it, but include my notes? [21:23] that seems reasonable. [21:23] jcsackett: I flipped that while tweaking it and it swapped things around [21:23] gary_poster: sure can do [21:23] thank you [21:24] jcsackett: do me a favor, can you put my face on that card and I'll peek at it when I get a chance. If that's all I need to do I can re-apply and make sure to add a better functional test a layer up to catch that stuff [21:26] rick_h: your mug is on it too now. :-P [21:38] gary_poster: if you wouldn't mind reviewing just to make sure I have your comment correct https://bugs.launchpad.net/juju-gui/+bug/1191160 [21:38] <_mup_> Bug #1191160: url shows /sidebar/ after deploying charm [21:39] that's it hatch, thanks. [22:07] gary_poster: thanks for the grammar fixes :) [22:07] English never was my strong suit ;) [22:08] hatch, lol, I was just trying to make it easier for me to parse [22:08] I missed at least one comma in there, and maybe put in an extra :-P [22:08] haha [22:37] hatch: picky :P (re url bug) [22:38] hatch: the trouble is that it's the url state, without doing an additional dispatch not sure we can get rid of it. [22:38] rick_h: lol, hey I just want the app to be perfect :) [22:38] hatch: hmm, maybe there's a way...anyway :P on the whole thing [22:38] where there's a will...there's a way! [22:38] mama always said, life is like a box of chocolates [22:38] oh wait...wrong quote [22:38] :P