| gary_poster | bcsaller_, if I qa and give you a LGTM can you land now? :-) | 00:17 |
|---|---|---|
| bcsaller_ | gary_poster: I think so | 00:17 |
| gary_poster | yay! ok 1 sec | 00:17 |
| gary_poster | bcsaller_, "Assuming you clean up review notes, LGTM! Yay, thanks!" | 00:23 |
| hatch | gary_poster: still around? | 00:26 |
| gary_poster | hatch, y | 00:26 |
| hatch | http://bazaar.launchpad.net/~hatch/juju-gui/promise-conflict/revision/646 mind trying this out whenever you get a chance to see if it fixes the loading screen issue | 00:26 |
| hatch | it solved it for ben and I | 00:26 |
| gary_poster | huh, weird | 00:27 |
| gary_poster | ok | 00:27 |
| hatch | yeah right? | 00:27 |
| hatch | heh... | 00:27 |
| gary_poster | hatch solves it. don't love the solution ;-) looking for a sec... | 00:30 |
| gary_poster | hatch, "firing "update" is | 00:35 |
| gary_poster | // unnecessary, and potentially even mildly harmful. | 00:35 |
| gary_poster | " | 00:35 |
| gary_poster | My current suspicion is that this is the problem | 00:36 |
| gary_poster | It gets the service | 00:36 |
| gary_poster | then gets the charm | 00:36 |
| gary_poster | fires update... | 00:36 |
| gary_poster | and we redispatch | 00:37 |
| gary_poster | meanwhile we've been loading haproxy because of the delta stream | 00:38 |
| gary_poster | which *also causes a redispatch | 00:38 |
| gary_poster | * | 00:38 |
| gary_poster | which then interrupts the connection of the promise and the view (actually it already did that; but the first time it was ok...) | 00:39 |
| gary_poster | so this is the third time through | 00:39 |
| gary_poster | (the update causes the three-alarm fire, at the very least) | 00:39 |
| hatch | hmm interesting | 00:40 |
| hatch | taking that fire out still causes it to render the three notifications but.... it looks like it's solved the hangup... | 00:42 |
| gary_poster | hatch, eh, I dunno, I should have kept it to myself. There's something unpleasant happening because of that update, but it's not the only problem; commenting it out... | 00:42 |
| gary_poster | right | 00:42 |
| hatch | ahh nope that didn't fix it | 00:42 |
| hatch | just got the loading screen | 00:43 |
| gary_poster | yeah not surprised :-/ | 00:43 |
| gary_poster | the update is not right, where it is, but not the problem | 00:43 |
| hatch | yeah it looks like it can be safely removed now | 00:43 |
| gary_poster | not sure about that | 00:44 |
| hatch | no? | 00:44 |
| gary_poster | we need to double check that the endpoint stuff still works in particular | 00:44 |
| gary_poster | that's where it came from | 00:44 |
| hatch | it appears to | 00:49 |
| hatch | at least as far as the ui interactions are concerned | 00:49 |
| gary_poster | hatch, huh. that would be cool to remove. since it doesn't seem to solve anything concrete atm, could you run that change past bac tomorrow and see if he has any qa ideas? | 00:50 |
| gary_poster | but does not need to be in this branch | 00:51 |
| hatch | sure - and thoughts on the 'fix' ? leave it in and create a follow-up ticket? | 00:52 |
| gary_poster | gimme another minute on it, but probably | 00:53 |
| gary_poster | the fact that we are loading the services/charms because of the delta stream at the same time as the unit view traversal code makes it confusing | 00:53 |
| gary_poster | the promise code is being called all over the place | 00:54 |
| gary_poster | the modelController stuff | 00:54 |
| gary_poster | mostly for deltas | 00:54 |
| hatch | yeah - it's going to take quite a bit of work to remove the dispatches entirely | 00:54 |
| gary_poster | and then three times for the view | 00:54 |
| hatch | but it's definitely doable | 00:54 |
| gary_poster | would you wait to dispatch until...what? | 00:55 |
| gary_poster | the delta has been processed and we have all models and charms? | 00:55 |
| gary_poster | oh, wait a second | 00:55 |
| gary_poster | I suggested a caching mechanism | 00:55 |
| gary_poster | not there | 00:55 |
| gary_poster | lemme hack that in, for the heck of it | 00:55 |
| gary_poster | well, lemme review the "then" docs again... | 00:56 |
| gary_poster | I turn into a pumpkin in 4 minutes | 00:56 |
| hatch | haha no problem | 00:58 |
| gary_poster | yeah you can call then multiple times on the same promise, as long as they return what you want the next "then" to have | 01:00 |
| gary_poster | aigh! turning...into...a...pumpkin! | 01:01 |
| gary_poster | hatch! go...ahead...and...land... | 01:01 |
| gary_poster | (||) | 01:01 |
| gary_poster | ^^^ pumpkin | 01:01 |
| hatch | lol nice pumpkin | 01:01 |
| gary_poster | not good ascii art, sorry | 01:01 |
| gary_poster | night! :-) | 01:01 |
| hatch | night | 01:02 |
| rick_h_ | hey huwshimi, how goes? | 01:23 |
| huwshimi | rick_h_: Hey. Pretty good. Trying to get through these bugs | 01:24 |
| rick_h_ | huwshimi: cool, just wanted to check in if you need anything | 01:24 |
| huwshimi | rick_h_: I don't think so at the moment. Thanks for landing all my branches. | 01:25 |
| rick_h_ | huwshimi: np, thanks for the update. Looks nice with the diff sizes/etc now | 01:25 |
| huwshimi | rick_h_: Yeah, much better :) | 01:25 |
| huwshimi | rick_h_: Things going OK at your end? | 01:27 |
| rick_h_ | huwshimi: we'll see. I'm out tomorrow so just working on getting the laptop setup on raring | 01:28 |
| huwshimi | rick_h_: OK cool, success? | 01:28 |
| rick_h_ | it's installed, still need to re-setup my environments for charmworld/juju-gui for working on stuff | 01:29 |
| huwshimi | Fun! | 01:29 |
| rick_h_ | party party | 01:30 |
| huwshimi | :) | 01:30 |
| rick_h_ | let me know if you need anything and I'll look out for any branches or at list ping others in the morning | 01:30 |
| huwshimi | rick_h_: Thanks, will do. Enjoy your evening :) | 01:30 |
| rick_h_ | hopefully we'll get the demo stuff all wrapped up | 01:30 |
| gary_poster | hatch, when you are around, http://paste.ubuntu.com/5629085/ is another way of addressing the redraw problem we saw (ignoring lines 56-59, and lines 87++). It is still a bit of a band-aid, but it is a good change IMO that happens to fix the immediate problem. Preventing the three-times dispatch will be the better fix later, but we'll still want this. | 13:18 |
| gary_poster | teknico, LGTM thanks | 13:22 |
| teknico | gary_poster, I'd do anything to make Sublime happy, you know ;-) | 13:23 |
| gary_poster | teknico, lol | 13:23 |
| rick_h_ | luca_: replied to your comments on the badges. If there's an issue please open a new bug to address how the .png should be updated. | 13:26 |
| frankban | gary_poster: morning. "Make selenium CI tests for sandbox". Is this switching to sandbox=true and then, e.g. deploying a charm? | 13:26 |
| luca_ | rick_h_: Ah, I wasn't sure what to do :) | 13:26 |
| gary_poster | frankban, yeah, that would be a good start :-) | 13:26 |
| rick_h_ | luca_: yea, it's all good. Since the change has landed/etc it's a new change to adjust it going forward. | 13:26 |
| frankban | gary_poster: cool | 13:27 |
| gary_poster | frankban, did you get a chance to look at charm tests? | 13:27 |
| frankban | gary_poster: yes, they passed locally two times in a row | 13:27 |
| gary_poster | frankban, ok cool, good enough. Thank you very much. I think it would be great to have an effort to see what needs to happen to make them reliable for everyone. I wish they weren't so expensive to work on. :-/ | 13:28 |
| gary_poster | We can talk about that next week | 13:28 |
| frankban | gary_poster: sounds good | 13:29 |
| gary_poster | cool | 13:29 |
| frankban | gary_poster: I see you also have a retrospective card re charm regression. less fragile tests and setting up a CI for the charm (including lint/pep8) could improve the situation IMHO | 13:31 |
| frankban | https://pypi.python.org/pypi/flake8 is interesting | 13:31 |
| gary_poster | frankban, +1. I think there's a discussion, and maybe a tough decision, about CI vs. tarmac also though. The issue with the charm is that when you push to trunk, it is *released*. We don't have a release mechanism like with the GUI itself. CI is really too late. Alternatively, maybe we set up a different trunk, or a different release branch, and release to it after CI and QA. That might work with other things | 13:33 |
| gary_poster | we need to do. | 13:33 |
| frankban | gary_poster: got it. I wonder if this could also be the right time to make deploy tests use juju-core | 13:38 |
| gary_poster | frankban, I think we need both. Which will make the tests even slower. :-/ | 13:39 |
| gary_poster | possibly also more difficult to arrange. | 13:39 |
| frankban | gary_poster: do you know if jitsu test is compatible with juju-core? | 13:39 |
| gary_poster | frankban, I don't know. However you are cordially invited to the meeting for this blueprint ;-) https://blueprints.launchpad.net/ubuntu/+spec/servercloud-s-juju-charm-testing | 13:41 |
| frankban | :-) | 13:41 |
| gary_poster | that will be sometime next week | 13:41 |
| * frankban thinks our own custom test runner could not be a bad idea [spoiler] | 13:42 | |
| gary_poster | heh | 13:44 |
| gary_poster | might be interested in that | 13:44 |
| hatch | gary_poster: thanks lookin gat diff | 13:59 |
| gary_poster | cool | 14:00 |
| sinzui | hatch, jcsackett, someone? Huw has a branch that needs another review https://codereview.appspot.com/9159044/ | 14:10 |
| gary_poster | sinzui, done | 14:15 |
| gary_poster | sinzui, is there a test deployment I should look for today? I got an impression from Rick that you might want me to update the demo site today? | 14:15 |
| sinzui | gary_poster, we are still evaluating that. We have a branch that will unblock a lot of work. I suppose it will be in the afternoon when we decide if we have something to show off | 14:21 |
| gary_poster | sinzui, ok. I'm hoping to take the afternoon off, but I can sneak in and see if you want an update | 14:22 |
| sinzui | okay. understood | 14:23 |
| hatch | I like your diff, couple questions... | 14:31 |
| gary_poster | sure | 14:31 |
| hatch | well just one | 14:31 |
| hatch | how does this solve the problem? heh | 14:31 |
| gary_poster | :-) | 14:31 |
| hatch | I like how we don't create new instances | 14:31 |
| hatch | but isnt' the execution the same? | 14:32 |
| gary_poster | hatch, first, tbh, this was a cleanup that I thought would solve another issue. It didn't. Then I noticed that this was cleaned up. bonus! ;-) That admission made, now you know I'm kinda guessing, but... | 14:33 |
| gary_poster | My hypothesis is that the fact that the "then"s are called synchronously makes the difference | 14:34 |
| gary_poster | So, old code: | 14:34 |
| gary_poster | promise 1: showView | 14:34 |
| gary_poster | [stuff happens] | 14:35 |
| gary_poster | promise 2: showView | 14:35 |
| gary_poster | [stuff happens] | 14:35 |
| gary_poster | promise 3: showView | 14:35 |
| gary_poster | New code: | 14:35 |
| gary_poster | promise: showView, showView, showView! | 14:35 |
| gary_poster | [stuff didn't get a chance to happen] | 14:35 |
| gary_poster | hatch I realize that's lame | 14:36 |
| gary_poster | but that's what I got ;-) | 14:36 |
| hatch | haha - well it's a hypothesis that fits | 14:36 |
| hatch | so I'll take it! | 14:36 |
| gary_poster | :-) | 14:36 |
| gary_poster | hey jujugui. I propose that we postpone our weekly retrospective until we see each other next week face to face. However, let's have our daily call in 9 minutes at the usual Friday time. please update kanban | 14:51 |
| Makyo | gary_poster, +1 | 14:53 |
| gary_poster | cool | 14:53 |
| hatch | I vote that we stop using paste.ubuntu.com in favour of something with syntax highlighting :) | 14:57 |
| gary_poster | it has syntax highlighting! you just have to select it... | 14:57 |
| hatch | ohhhh so it's allyalls fault LO | 14:58 |
| hatch | :P | 14:58 |
| Makyo | Pff. | 15:00 |
| gary_poster | jujgui call now | 15:00 |
| gary_poster | sorry | 15:00 |
| gary_poster | jujugui :-P | 15:00 |
| * gary_poster signed up for Selenium for Pythonistas. Thanks bcsaller_ :-) | 15:26 | |
| bcsaller_ | cool :) | 15:26 |
| gary_poster | oops | 15:26 |
| hatch | there are a number of javascript meetups but they are all in SF | 15:29 |
| gary_poster | bcsaller_, fwiw landscape is working fine with improv. yay! so, no worries. | 15:30 |
| bcsaller_ | gary_poster: ha, cool, I'll check what I didn't set up properly | 15:31 |
| gary_poster | cool | 15:31 |
| Makyo | Back in a little bit. Getting a tour since it's the first day. | 15:31 |
| bcsaller_ | gary_poster: (and anyone else) if there are any mutations you'd like to see let me know | 15:31 |
| gary_poster | bcsaller_, just to screw with people we could move the services around the screen ;-) | 15:32 |
| gary_poster | *note that was not a suggestion ;-) | 15:32 |
| bcsaller_ | gary_poster: already started, I have one that should occasionally flip/mirror the setup | 15:32 |
| gary_poster | heh | 15:33 |
| gary_poster | cool | 15:33 |
| hatch | bcsaller_: any chance you could make one of your little simulators that could simulate a slow back end? | 15:33 |
| hatch | so we can test for race conditions and the like | 15:33 |
| bcsaller_ | hatch: that would maybe be something else. This creates mutations, we could add somethign at the sandbox layer that delayed outoing responses, but I'm not really sure how useful that is | 15:34 |
| hatch | well right now with rapi and fakebackend everything returns really fast | 15:36 |
| hatch | I'd like to see what happens if there is a hangup somewhere | 15:36 |
| bcsaller_ | hatch: but you'd want that relative to user-generated events, the stuff coming out of simulator is random and acts as though another use made remote changes to the env | 15:39 |
| bcsaller_ | so the delay would be unknowable | 15:39 |
| bcsaller_ | but a delay slider relative to sandbox replies is do-able outside this effort | 15:39 |
| hatch | ahh I see what you're saying | 15:40 |
| hatch | bcsaller_: are you here today? | 16:09 |
| bcsaller_ | hatch: yes | 16:11 |
| hatch | ok great, I am just fixing the tests from garys patch to that race condition and then I'm going to propose this so I'd like another QA just to be safe | 16:11 |
| bcsaller_ | hatch: let me know | 16:14 |
| hatch | just finished the fixes so now running the suites again it'll probably be 10-15 | 16:14 |
| gary_poster | jujugui, stepping away for afternoon. Will occasionally check back. I have a lot of work to do before Monday, so if you want to get in touch, I will be around eventually. :-) I look forward to seeing everyone soon! | 16:21 |
| hatch | enjoy! see you Tuesday | 16:22 |
| bcsaller_ | gary_poster: have a good one | 16:22 |
| gary_poster | thanks :-) | 16:22 |
| hatch | bcsaller_ if you could also plz qa this https://codereview.appspot.com/9178043/ | 16:22 |
| hatch | we can say gary and I paired so I'll land after one review :) | 16:23 |
| * hatch is a rebel with a cause | 16:23 | |
| Makyo | Oops, reboot. | 16:25 |
| hatch | bcsaller_: if I could bug ya to qa that that would be awesome :) I'd like to finally drag that card into landed :) | 16:49 |
| bcsaller_ | hatch: sorry I didn't say something, I've been working on it | 16:49 |
| bcsaller_ | hatch: still get the 3+ notifications, you didn't want to add the timer around that? | 16:49 |
| hatch | I do....but I'm a little worried that it will then hide the real problem | 16:51 |
| bcsaller_ | hatch: The simulator aggressively shows refresh 'opportunities' , but doing a plugin that forces semi-random traversals might be fun | 16:55 |
| hatch | haha oh boy | 16:57 |
| hatch | alllright finally merged that in! | 17:11 |
| hatch | bcsaller_, it looks like the CI failure in IE was from your branch... for the sandbox import...IE probably doesn't support the same methods | 17:26 |
| bcsaller_ | nothing should touch the html5 codepaths | 17:26 |
| bcsaller_ | I can test my current stuff on ie soon though | 17:28 |
| hatch | alright thanks - it just looks like that's when the latest failures started | 17:41 |
| hatch | or it could be something entirely unrelated of curse :) | 17:41 |
| hatch | bcsaller_, fakebackend.importEnvironment successfully imports v0 data - timeout exceeded | 17:43 |
| hatch | so it probably just takes too long | 17:43 |
| bcsaller_ | hatch: thats possible, I'm seeing errors in BrowserCharm test on IE, but I'll up the timeout on import | 17:53 |
| hatch | hmm those aren't showing up in the CI video | 17:54 |
| hatch | /join #yui | 18:01 |
| hatch | /join #yui | 18:01 |
| hatch | ... | 18:01 |
| Makyo | Oh nooooo. Working in the precise container broke lbox D: | 19:27 |
| rick_h_ | gary_poster: sinzui pushed up a working fix to the ~juju-gui/browser-default branch | 19:39 |
| hatch | bcsaller_, the other day I picked up FTL - I haven't been able to get more than 1/2 way on normal haha | 20:04 |
| bcsaller_ | hatch: yeah, that one has a steep learning curve | 20:04 |
| hatch | I got to the end on easy and got the final boss down to about 1/4 but that's it | 20:05 |
| hatch | I never have any money so my ship always sucks | 20:05 |
| bcsaller_ | hatch: I bet you clear zones too fast then | 20:06 |
| hatch | yeah? I've tried but maybe I need to keep the fleet RIGHT behind me | 20:07 |
| hatch | I'm guessing that you have been more successful? | 20:08 |
| bcsaller_ | hatch: yeah, I don't play much now, but I do have all the ships | 20:17 |
| hatch | heh I have 3 :) | 20:18 |
| hatch | I have been looking at Kerbal Space Program game | 20:18 |
| hatch | looks pretty cool | 20:18 |
| hatch | http://store.steampowered.com/app/220200/ | 20:19 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!