gary_posterbcsaller, are you safe in SF?12:51
gary_posterif he is it would be very early there...12:52
rick_h_heh, zzzzzz12:52
* rick_h_ goes to check for SF quakes or the like12:52
bachi gary_poster13:18
gary_posterhey bac13:18
bachey i'm having some issues dealing with all of the legacy test failures my branch introduced.  some are quite odd, including trying to create charms that claim to have no id, despite clearly passing one.13:19
bacgary_poster: do you have time to chat and maybe give me some ideas?13:19
gary_posterbac, sure.  should I get a copy of branch?13:20
bacyeah, lp:~bac/juju-gui/get-endpoints13:20
baclet me push some debugging changes13:20
bacok, grab now13:21
bacgary_poster: in guichat13:22
gary_posterack, there soon13:22
gary_posterhey hatch, when you start, assuming you don't have anything in progress, why doncha pair with bac13:37
bacgary_poster: oh, i meant to mention, to do some of that debugging i had to extrace a variable declaration and assignment out of the big block of them in order for the debugger to not skip over it.  an argument (too late) for 'one var per declaration'.13:38
bacammo for the next time it comes up.  :)13:39
gary_posterbac, oh, interesting!  maybe microblog it--what you just said on IRC-- on the blog so you don't forget?  up to you13:39
hatchsure thing13:57
rick_h_hatch: do you have details around the _yuid changing on a node? I tried to do some searching this weeekend but didn't find anything 13:58
rick_h_hatch: is there some condition that arises? Or it's just unreliable in general?13:58
hatchrick_h_: well the yuid doesn't change on the same node, but if it gets re-rendered then it will be different13:58
rick_h_hatch: gotcha, but that should be ok as you target an indicator at a container node while you update it's contents. 13:59
rick_h_otherwise your indicator would vanish 13:59
hatchyeah what I meant was that if your indicator node gets removed for whatever reason - then you try and show it13:59
hatchit'll show on a node that's not in the dom any longer13:59
rick_h_hatch: right, but the point of this is to make sure the indicator is removed when a view is destroyed/the node goes away14:00
rick_h_so for the lifetime of the view it should be good on the container, then the veiw goes away and all indicators used are wiped up14:01
hatchoh yes right - It would be more of a concern if it was more general use14:02
rick_h_hatch: ok, yea. The point of this is just an extension to use on views to help make sure no indicator is left behind14:02
rick_h_since it'll be used on a bunch of browser views. 14:02
rick_h_hatch: I can add docs to help narrow the written purpose if that helps. Aside from that can you give it a review look over sometime please?14:03
hatchoh yeah for sure14:03
rick_h_hatch: cool thanks. Appreciate it14:03
hatchI'm just gona grab a coffee then I'll get right on it14:03
rick_h_hatch: cool https://codereview.appspot.com/8186045/ to save you link hunting14:04
hatchbac: are you still having issues?14:21
bachatch: working through them14:21
hatchalright, did you need me to take a look?14:21
bachatch: if you have nothing else to do we could pair after the call.  otherwise i could just keep going and maybe bug you if i get stuck14:22
hatchoh I have other stuff but could take a look if you need14:23
hatchso I guess lemme know14:23
hatchbac: a comment on your blog post - you could put the breakpoint in the makeBar function :)14:25
bachatch: you could.  but what if makeBar is called in lots of other situations that are not interesting?14:26
hatchthen life sux!14:27
hatchlol you make a good point though14:27
hatchI haven't actually ran into that before14:27
MakyoI have, but I've just used F10 to step over to the line I need.14:28
rick_h_yea, I just do the debugger before the var declarations and run the function from the console 14:28
bacMakyo: really?  when i tried 'step over' it went to the next line out of the declaration14:28
rick_h_the only time it's a really big pain is if it's foo=something(), bar=morestuff(foo)14:29
hatchrick_h_: yeah I don't like that layout14:29
Makyobac, Hm.  Let me make sure I'm not lying.  This might've been a chained thing instead.14:29
rick_h_in which case I probably end up breaking the var up14:29
MakyoOops.  jujugui call now.14:30
gary_posteroops thank you14:30
hatchO K now to see if I remember how this add relation code worked14:39
gary_posteradd unit?14:40
hatchyeah...see I even forgot what it's called14:40
hatchso how was everyones Easter? Eat too much food?14:44
gary_posterkids go crazy with too much sugar14:47
gary_posterbac, do you know diff between []params.Port(nil) and []params.Port{} ?  Test expects first and obtains second14:47
hatchI was thinking when I have kids I'll hide carrots and lettuce - afterall they are from a rabbit14:47
MakyoBathed dogs, who then ran around at the dog park and got all gross again.14:48
hatchMakyo: yeah - it's impossiblet o keep dogs clean in the spring14:48
* hatch clearly forgot how to type too14:48
rick_h_labs ftw14:48
hatchrick_h_: are you saying labs don't get dirty?14:49
bacgary_poster: i don't.14:49
gary_posterk thanks14:49
rick_h_hatch: yea, their coats are super easy thankfully14:50
bacthe second is obvious but i'm not familiar with the first14:50
MakyoThankfully both the sheps have short coats, so they clean up easily.  14:50
hatchrick_h_: oh haha - my dogs are only like 7" off the ground so they get dirty regardless :)14:50
hatchthe app.js initializer is almost 200 lines...I think it needs to be split up hehe14:54
rick_h_hatch: we have to go back to this multiple call business still as well. It's starting to get annoying heh14:56
hatchmultiple dispatch?14:57
rick_h_hatch: yea14:57
* hatch lights the torches and hands one to rick_h_14:57
rick_h_since I'm scolling down to view my work it keeps moving the windows focus on me 14:57
hatchwhy don't you set your container to position absolute? :)14:58
* rick_h_ loads weapon of CSS destruction14:58
hatchhaha it shouldn't break anything14:59
hatchalthough you'll probably have to set position relative on the container15:00
hatchwhich might break everything15:00
hatchohh funny story15:02
hatchI stepped through my garage roof yesterday15:03
hatchyesterday - a day15:03
hatchheh woops15:03
bachey our wordpress blog has a fan. ronscubadiver from TX just like my post. we have an audience!15:04
hatchgary_poster: are we using _'s in sandbox.js?15:05
gary_posterhatch, only for legacy bits15:05
gary_posterhatch which means the handlers15:05
hatchalright gotcha - I was kind of wondering if these shoudln't be in a closure so we didn't have to prefix them15:06
hatchit doesnt really matter15:06
hatchjust thinking :)15:06
hatchjujugui - uistage is broken a conflicted file was comitted15:25
gary_posterhatch fixing not committed I don't think checking15:25
* bac looks15:25
bacgary_poster: you doing it? or me?15:25
gary_posterbac you thank you15:25
gary_posterbac lplease lemme know when done15:25
hatchgary_poster: am I correct that I can't test this add_unit code until I first implement get_service?15:29
gary_posterhatch, no.  you need deploy, which you already have15:29
gary_posterhatch, deploy via the fake backend, get the service from the result, and then use that id for the add_unit15:30
gary_posteryou can actually provide the id explicitly and then you don't even need to look at the result15:30
hatchoh ok I need to manually trigger it15:30
hatchI was trying to figure out how add_unit was called from the UI15:31
gary_posterwell, since you are not testing that part of the API, that makes sense to me15:31
gary_posterthat's the test prep15:31
bacgary_poster: uistage happy again15:34
gary_postercool thanks bac15:34
gary_posterrick_h_, ^^15:34
rick_h_gary_poster: thanks15:34
gary_posterwe really should deploy that from a charm 15:34
gary_posterrick_h_, do you have a deployment plan for jujucharms.com, he said with a shudder?15:35
rick_h_gary_poster: otp, sec15:36
rick_h_gary_poster: so, from what I understand it's been handed off to #is to complete getting it deployed to prodstack. 15:46
rick_h_gary_poster: sinzui will know more/status than I at the moment ^15:46
gary_posterrick_h_, but I meant the GUI itself15:47
rick_h_gary_poster: oh, the deployment story I imagine is getting the gui charm up and running on prodstack under that domain. 15:47
gary_posterrick_h_, and that won't be *any problem at all* will it ;-)15:47
rick_h_gary_poster: I've got nothing about deploying gui than you guys have. I'm jumping on your train15:47
rick_h_sorry, assumed it was a yellow problem and such :)15:48
rick_h_gary_poster: well, honestly shuoldn't be since you guys don't have external services to HA and the like. The only thing will be gettnig HA proxy in front of the Gui and making sure multiple instances can run side by side. 15:48
gary_posterrick_h_, ok, cool.  I had the "jujucharms.com is for orange" in my head.  glad we are ironing out.  that could have been a  nasty surprise15:48
sinzuiNo movement on our ticket. I did update it make it clear we have delivered our part for manage.jujucharms.com, and that it is required to deliver juju-gui  as jujucharms.com15:49
rick_h_gary_poster: definitely. Orange is moving jujucharms.com to management.jujucharms.com which is our charmworld pyramid app on prodstack15:49
rick_h_gary_poster: jujucharms.com should be the prodstack deployed juju gui charm15:49
gary_posteryeah, in our configuratuion they shouldn't even need to talk to one another: it will be in-browser "juju"15:49
rick_h_gary_poster: maybe you can sinzui and the people working on the gui charm should meet up and head off any craziness at the pass15:49
gary_posterso, rick_h_, the only challenge I know of will be that they will want to be able to get the tarball and any deb dependencies from their own locations15:50
bachatch: can you confirm that browser_charm_view fails on trunk for you?15:50
gary_posterrick_h_, +115:50
bacit didn't look like anything i would've broken15:50
gary_posterbac I had to make clean before it passed15:50
* bac tries15:50
gary_posternot ideal but I didn't have time to investimagate15:50
baci'm glad to know 0 does not deeply equal 215:50
MakyoOut of control notifications.  Back in a sec.15:50
hatchbac sure15:51
* hatch switches gears15:51
bachatch: nm, given gary_poster's report15:51
* hatch switches back15:51
gary_postersinzui, do you think depoyment is good topic for call in 8 minutes with group?15:52
gary_posterdeployment of GUI on prodstack for jujucharms15:52
sinzuigary_poster, yes15:52
gary_posterok cool15:52
bacthanks gary_poster.  wfm now15:52
gary_postercool bac15:52
sinzuigary_poster, as I hinted in the last call. We are waiting for manage.jujucharms.com to be deployed. It is a requirement for juju-gui to become jujucharms.com15:53
gary_postersinzui, right.  I didn't think about the fact that the GUI charm itself will need to be hardened for prodstack15:53
gary_posterI *think* I know what that will entail15:53
gary_posterand for our purposes I don't think it will be too much15:54
gary_posterjust make it possible to get tarball and debs from custom locations15:54
gary_posterand not reach out to other locations in that case15:54
rick_h_arosales: so with staging happy http://uistage.jujucharms.com:8080/bws/sidebar should load. Let it load then scroll down below the environment. 16:12
rick_h_arosales: click on the charm to view the details, same thing. Scroll down to see the start of display/tabs. No UX so it's just framing for the JS view/functionality vs look/feel :)16:13
arosalesrick_h_: thanks. I'll take a look now . .  .16:20
hatchrick_h_: pretty cool :)16:20
hatchhurry up and fix so I can demo the router stuff :P16:21
hatchgary_poster: in the tests I have state.deploy('cs:wordpress', function() {}); and then before that env.get('conn').onmessage(function(received) { ... }); but that callback is never triggered16:23
hatcham I missing how this works?16:24
hatchlet me rephrase that16:24
hatchI am missing how this works16:24
gary_posterhatch sorry calls16:40
hatchnp I may have it16:40
hatchtesting now16:41
gary_posterok cool hatch.16:41
hatchgary_poster: still on a call? https://code.launchpad.net/~hatch/juju-gui/add-unit my test fails in mocha but passes in the browser17:00
hatchTypeError: JSON.stringify cannot serialize cyclic structures. (mocha-phanto│GET /juju-ui/assets/images/picker_1px.png 200 2ms - 96517:00
hatchit's also very slow +1s17:01
gary_posterno on call hatch. you mean fails in phantom right?17:01
hatchyeah...oh man I need another coffee or something hah17:01
hatchI'm assuming I"m doing this wrong becuase there is no way that tests should take over 1s17:01
gary_posterverified symptoms hatch.  now looking at code17:05
hatchmy experience with the environment code is virtually 0 so I woudln't doubt I'm doing this wrong17:06
gary_posterhatch, let's guichat17:08
=== matsubara is now known as matsubara-lunch
=== Makyo is now known as Makyo|out
Makyo|outHaircut time17:48
hatchgary_poster: http://pastebin.ubuntu.com/5668110/17:54
hatchany input?17:54
gary_posterhatch, #1, clear out the delta stream, as we said17:55
gary_posterbefore you open17:55
hatchoh right - I had that in there but didn't appear to make any difference17:56
hatchI'll add it back17:56
gary_posterhatch #2, the fact that you are waiting 1 second means that you are still not returning an initial response (client.receive, like in the login op handler)17:56
gary_posterbecause waiting one second means that you are waiting for the delta stream17:57
gary_posterwhen you should be waiting for the client response17:57
hatchahh ok17:57
hatchoh looks like I had a typo there17:57
hatchman we REALLY need to document attrs and events18:12
gary_posterhatch, write down the pain, because otherwise you will forget it18:13
gary_posterthat is the specific things that you mean18:13
hatchyep for sure18:13
=== matsubara-lunch is now known as matsubara
bachatch: when you have some time i'd like to get some help18:21
hatchsure 5mins?18:21
hatchI have new neighbours today....wonder what they will be like18:22
* hatch hopes they don't throw huge parties every weekend18:22
gary_posterhatch, interestingly the unit tests were fine, without any retries, for two or three times in a row.  The most recent ran the unit tests four times before it got a success, but at least it did eventually get a success. :-/
hatchman that's so irritating18:27
hatchit never fails in chrome though does it?18:27
hatchbac: ok I'm at a good break point18:27
hatchwhat's up?18:27
=== BradCrittenden is now known as bac__
=== bac__ is now known as bac
ahasenackhi guys,18:47
ahasenackI deployed juju-gui using "stable", and /var/lib/juju/units/juju-gui-0/charm/juju-gui/build-prod/juju-ui/version.js confirms that18:47
ahasenackvar jujuGuiVersionInfo=['0.3.1', '470'];18:47
ahasenackbut I don't see a "landscape" link in the lower right corner, it's been a while since I last deployed this18:47
ahasenackmaybe I'm just not finding it, where should it be, does someone know?18:48
gary_posterahasenack, the link does not show up until landscape has annotated the environment18:48
ahasenackah, ok18:48
ahasenackso I have some bug hunting to do :)18:48
hatchgary_poster: I think bac may have found our test failing issue18:53
hatchwell whatever he is doing is causing the tests in application basics to fail18:53
hatchbut randomly18:53
hatchsometimes it passes18:53
gary_poster:-) fun18:53
hatchsometimes 2 fail, sometimes 8 fail18:53
gary_postersounds different to me18:53
hatchsure - but if you look at the test suite you'll notice that it doesn't clean up after itself18:57
ahasenackgary_poster: what would be a line in the logs that confirms landscape is sending the annotations? I have seen some like these:18:58
gary_posterhatch, if we fix it, I'd be thrilled :-)18:58
ahasenack2013-04-01 14:48:40,793: juju.rapi.ws@DEBUG: send delta update [('service', 'change', {'charm': 'local:precise/keystone-213', 'annotations': {u'landscape-computers': u'%2Bservice%3Akeystone', u'gui-y': 153.5838711310003, u'gui-x': 570.5940154188667}, 'id': 'keystone'}), ('service', 'change', {'charm': 'local:precise/nova-compute-87', 'annotations': {u'landscape-computers': u'%2Bservice%3Anova-compute', u'gui-y': 277.3118:58
ahasenack58501443403, u'gui-x': 869.4793787934253}, 'id': 'nova-compute'})]18:58
ahasenack2013-04-01 14:48:25,691: juju.rapi.ws@INFO: Process message {"op":"update_annotations","entity":"keystone","data":{"gui-x":570.5940154188667,"gui-y":153.5838711310003},"request_id":3}18:59
ahasenackbut I'm not sure who sent them18:59
ahasenackother than that, no easy traceback on either side so far18:59
ahasenackstill digging18:59
gary_posterahasenack, the second one is unrelated18:59
gary_posterthat is the gui annotating viaula location of a service19:00
gary_posterthe first is promising19:00
ahasenackyeah, figured (x-y coordinates)19:00
hatchviaula sounds like some medication name ;)19:00
ahasenackyeah, that first one has bits of a valid landscape link19:01
ahasenackthe service:nova-compute part19:01
ahasenackit's a landscape computer search19:01
gary_posterahasenack, if I had to guess, I would guess that the environment annotations are not being set/sent to the GUI19:01
=== Makyo|out is now known as Makyo
gary_posterthat provides the base of the URL19:01
gary_posterand it is what we use to determine if we should turn on the environment annotations19:01
gary_posterahasenack, ideally you would see these annotations in that stream.  maybe look for the string "environment"?19:04
gary_posterahasenack, other thinsg I can do to help, though it would take me a few minutes:19:04
ahasenackI can see landscape connecting to the api19:05
gary_poster- give you something to run in the browser to look at the annotations the the GUI knows about for the environment19:05
ahasenackI restarted that landscape service and I see the login op, and the api sending back a bunch of stuff19:05
ahasenackI need to watch for the other way around now19:05
gary_poster- give you something to run in the browser to ask Juju directly for annotations on the environment19:05
ahasenackin the "send delta update" bit, there are annotations in it19:06
gary_posterahasenack, when it sent a bunch of struff did it include "environment" associated with an update_annotations op?19:06
ahasenack'annotations': {u'landscape-computer': u'%2Bunit%3Aswift-proxy%252F1', u'landscape-needs-reboot': True}19:06
ahasenackI'll have to check, it's a long message19:06
gary_posteryeah, we established that services are getting annotations19:06
gary_posteryeah, grep it! :-)19:07
ahasenackno update_annotation yet19:08
ahasenackor "environment"19:08
* hatch goes to eat19:08
gary_posterI'm trying to verify on my side with test data19:08
hatchding if ya need me19:08
ahasenackolder bits of the logs do have it19:09
ahasenackhm, wait19:09
ahasenackthat's the x-y position ones19:09
gary_poster1 sec ahasenack I got something else for you to look for19:10
bacanyone seen google maps today?19:10
ahasenackthe only "environment" string I have is19:10
ahasenack# grep environment api-agent.log 19:10
ahasenack2013-04-01 11:44:01,019: juju.agents.api@DEBUG: Received environment data.19:10
gary_posterahasenack, I'm on crack, but I'm trying to wean myself off. ;-) see http://pastebin.ubuntu.com/5668359/19:11
gary_posterahasenack, those values are old19:11
gary_posterahasenack, but the keys are correct19:11
ahasenackgary_poster: so, I should look for a log entry similar to that one, with those keys19:11
gary_posterin particular we have to have landscape-url, but all those keys are necessary19:11
gary_posterprecisely ahasenack 19:12
ahasenackyeah, no landscape-url anywhere19:12
gary_posterahasenack, I'm suspicious that landscape is not giving us the keys we need anymore, then19:14
gary_posterwell, even more suspicious, since I had already qa'd this on our end ;-)19:15
ahasenackgary_poster: ok, I'll file a bug on our side19:15
gary_posterthanks ahasenack 19:15
ahasenackgary_poster: yeah, it changed to fix a bug19:15
ahasenackwouldn't be the first time a bug was introduced by a fix19:15
gary_poster:-) no19:15
bacgary_poster: vacation request submitted.19:20
rick_h_jcsackett: hatch if you guys get a few spare minutes :) https://codereview.appspot.com/817104719:34
hatchrick_h_: is there a risk of that qa.json becoming stale?19:40
rick_h_hatch: yea, but kind of the case with any sample data :/19:46
hatchyeah true true19:46
hatchMakyo: don't like your hair?19:56
Makyohatch, Nah, it came out fine.  Don't like my wireless router, though.19:57
hatchMakyo: your new hair? http://1.bp.blogspot.com/-bJLQi7wKPU8/T7EH46gZ_gI/AAAAAAAABNU/eRblcqMK6GU/s400/09-mohawk.jpg19:57
Makyohatch, I don't think I could quite pull that off.. :)19:58
hatchhow will you know unless you try?19:58
hatchyou miss 100% of the shots you don't take :P19:58
MakyoWho's to say I haven't?19:58
hatchtrue true19:58
Makyo(ps: I haven't)19:59
hatchgary_poster: in python.js ln 133 there is a line this.ws.send(msg) in _send_rpc() can you tell me where this is set?20:08
gary_posterlooking hatch20:09
hatchoh I found it20:10
gary_poster:-) k20:10
hatchbasically I'm trying to track down an error with env.add_unit()20:10
hatchon the integration test20:11
hatchhey gary_poster you wrote the addUnit method in fakebackend.js right?20:20
hatchwait ignore that20:20
hatchthis is my screwup20:20
hatchok lets roll that back just a bit20:23
hatchgary_poster: is it possible that the addUnit method doesn't return the proper formatted data? Or is it intentional that I'm supposed to match it to the real output?20:34
gary_posterhatch, there's only confidence if there is a test. :-)  Let's see if there is a test for addUnit in the test_fakebackend file...20:35
hatchwell it works20:36
gary_posterhatch, tests seem reasonable.  if you think there's a problem, maybe try to write a breaking one?20:36
hatchit's output just doesn't match the native one20:36
gary_posterhatch, oh!  no, that's the job of what you are writing20:36
gary_posterremember I said there was deterctive work involved?20:37
gary_posterdetective, even20:37
hatchyep ok - so then I'll break these tests20:37
hatchand that's ok?20:37
gary_posterhatch, um?20:37
gary_posterhatch, let's have a guichat :-)20:37
hatchsure :)20:37
jcsacketthatch, Makyo, rick_h_: can two of you fine folks look at https://codereview.appspot.com/807104420:43
MakyoSure, on it.20:43
jcsackettMakyo: thanks.20:43
gary_posterMakyo LGTM20:49
Makyogary_poster, thanks21:04
* Makyo walks dogs before it rains.21:04
gary_poster:-) welcome21:04
hatchgary_poster: when you get a second could you check out https://codereview.appspot.com/8236043/21:05
hatchI am pretty sure I got it this time21:05
* hatch has a thick skull sometimes21:05
gary_posterhatch, yeah.  Still a bit of trouble, I think.21:05
gary_postersorry, hatch, but let's guichat21:05
hatchgary_poster: code has been updated22:34
gary_posterhatch, just saw, looking22:35
* hatch crosses fingers22:36
gary_posterlooks good hatch, doing silly normal review things :-)22:36
gary_posterwill eventually result in a LGTM :-)22:37
gary_posterhatch, LGTM with small suggestion22:38
hatchthanks - sure I'll add that :)22:38
hatchI actually didn't know there was an isUndefined22:40
gary_posterassert is a nice interface.  I didn't start using it till I saw benji use it22:40
* hatch wonders what assertion lib we are actually using22:40
gary_posterat least that doc has worked well for me ;-)22:41
hatchhaha perfect22:41
hatchhmm after I added assert(isUndefined(data.err)); I got a 'verify dependency' error22:44
hatchbut now it's working just fine22:44
hatchthere is definitely some leakage in some tests here22:44
hatchmaybe after 13.04 we take a day and go through all of the suites to clean them up22:44
hatchi mean22:45
hatchbcsaller: able to take a look at my fakebackend branch for review?22:47
bcsallerhatch: got the link22:49
hatchgary_poster: if you're still around - any idea which task you want me to tackle next so I can do some evening reading22:56
hatchbcsaller: I just copied that signature from a previous test23:14
hatchoh it looks like it's from python.js ln 22723:15
gary_posterhey hatch.  I'm here for a sec.  you still here?23:50
gary_posterthanks for getting that branch landed. cool.23:50
hatchI am yep23:50
hatchI'm glad it's over :D23:50
hatchalthough I have a much much better understanding of what's going on23:50
gary_posterfirst smiley: fat finger led to fat lip23:52
gary_posterhatch, how about remove_units?  it's all in a similar part of the world, so it would be a natural change.23:59

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!