/srv/irclogs.ubuntu.com/2012/12/20/#juju-gui.txt

=== mattuk1972_ is now known as mattuk1972
=== _mup__ is now known as _mup_
dguerrithis might not be the right channel… but.. is there some docs about the juju RAPI?11:45
teknicodguerri, hi11:52
dguerrihi, teknico 11:54
teknicodguerri, not that I know of11:55
frankbandguerri: you could find some info on setting up the rapi-rollup agent in the documentation included in juju-gui: http://bazaar.launchpad.net/~juju-gui/juju-gui/trunk/view/head:/HACKING11:58
dguerrithanks frankban but I'm thinking about writing a ruby binding for the juju restful API12:03
frankbandguerri: it's a websocket API, it makes use of good rest principles, but still it's not a rest API. Unfortunately, my I am not aware of a place where the stream is documented.12:09
dguerrimmmh, so it might not be the right interface for a lang binding … 12:10
bacmorning frankban.  how goes the effort?12:12
frankbanbac: it's under review, you can glance at https://codereview.appspot.com/6977043/ . Tests results are different based on who runs the suite.12:19
bacfrankban: oh, that's good.  keeps people on their toes12:20
frankbandguerri: I guess it depends on what you want to achieve12:20
frankbanbac: :-)12:20
bacfrankban: is lp2kanban busted?  i see your defect in done-done has a bug but it isn't populated12:23
frankbanbac: it seems so12:23
frankbanbac: the same for other two cards in "ready to code"12:25
bacyep.  i'll look into it12:25
* frankban lunches12:38
bacfrankban: lp2kanban is undead again13:06
gary_posterhey benji.  I thought a fun improvement I could tackle today in very limited time might be to update the release docs a bit.  if that sounds ok to you and you are available to talk through things with me so I can pick your brain sometime, lemme know13:11
benjigary_poster: sounds like a plan13:11
gary_postercool13:11
gary_posterty13:12
gary_posterbcsaller, did you see what I meant about the rietveld review?  In a bit I'll make my own diff13:16
frankbanbac: cool thanks13:17
bacit is not very robust wrt ill configured boards and projects13:17
gary_posterbac, what cam we improve for poor configuration13:18
gary_posterthank you for fixing btw13:18
bacgary_poster: i just added some exception handling around the call to the board processing.  so if someone (<cough> green) isn't configured their processing stops but other boards continue13:18
gary_posterheh13:19
gary_posterok cool thanks bac.  you shot a note to gmb too?13:19
baci've pinged him, waiting for him to return from lunch13:19
gary_postercool13:19
bacbrits and their ill-timed lunches13:20
gary_posterEuropeans and their ill-timed lunches!  Americans and their ill-timed desire to sleep through till the morning!13:20
gary_poster:-)13:20
benjiHumans and their ill-timed biological needs.13:24
bcsallergary_poster: I was able to fix it, still don't know why it happened before 13:54
gary_posterbcsaller, cool, reviewing, thanks.13:54
gary_posterteknico, you are going to start bug 1074413, I'm guessing14:01
_mup_Bug #1074413: GUI charm should require HTTPS connections to the environment by default <deploy-story> <juju-gui:Triaged> < https://launchpad.net/bugs/1074413 >14:01
gary_posterbac, I made room for you on secondary story to start tests ofthe pan/zoom module, based on bcsaller14:02
gary_poster's branch in review14:02
gary_posterhazmat if you could land your branch that would really help us keep the flow moving14:02
hazmatgary_poster, eta 1hr14:04
gary_posterthanks14:04
teknicogary_poster, yep14:09
bacgary_poster: ok14:10
bacgary_poster: there is a card for bug 1083935 but it is blocked by 1077050 and 1077051.  only the latter seems to be fixed.  unblock it and carry on?14:18
_mup_Bug #1083935: Environment pan/zoom code is not tested sufficiently <juju-gui:Triaged> < https://launchpad.net/bugs/1083935 >14:18
gary_posterbac, yes, that's based on plans that have been superceded.  You are still blocked on Ben's branch, but since you are working from it, it's ok14:34
gary_posterbcsaller, quick checkin?14:34
bcsalleryeah14:34
bachi bcsaller, i just grabbed your current branch.  when i try to run the tests i get a time out on the first test and it stops.  anything different about running tests with your new work?14:45
bcsallerbac; left a .only on one of the test that _is_ having issues, reproposed the branch w/o that , sorry14:47
bacbcsaller: np, will grab new version14:48
bacbcsaller: cool, it now works except for that one test failure14:50
frankbangary_poster: hum... it seems that our debug server does not contain everything needed. I see this in the js console after switching to staging in a deployed charm: http://pastebin.ubuntu.com/1452506/15:11
gary_posterfrankban, ack on call will look soon15:12
frankbangary_poster: thanks, and... problems also in the production server: http://pastebin.ubuntu.com/1452512/15:16
gary_posterteknico, ^^^ ?15:16
gary_posterwill look asap15:17
teknicogary_poster, looking15:17
teknicofrankban, so, what files are missing?15:20
frankbanteknico: I don't know, my understanding is that https, refusing to load external insecure resources, exposes a problem we have: our releases don't contain everything required by the GUI, and the YUI loader tries to fetch libraries from the net (e.g. ellipsis, markdown, timer)15:24
teknicoah, I see15:24
gary_postersorry I didn't QA enough to catch that15:25
gary_posterteknico, previous call ran over, sorry.  talk after standup?15:25
teknicogary_poster, sure15:25
gary_posterfrankban, teknico we may need to revert the https code to let frankban proceed :-/15:25
teknicogary_poster, right. I still did not manage to see tests complete successfully :-/15:26
gary_posterteknico, I am afraid that this is going to be a very serious problem 15:28
gary_posterbac bcsaller benji frankban goodspud hazmat Makyo teknico call ni 215:28
frankbangary_poster: I am not blocked, and I already merged trunk and resolved conflicts. And, AFAICT, this is a bug of the GUI, not of the charm. However, is uistage broken?15:28
gary_posterfrankban, ui stage is not broken15:28
gary_posterbecause not served from charm, or from httos15:29
gary_posterhttps15:29
gary_posterbut this means that the charm is now unusable15:29
frankbangary_poster: yes15:29
gary_posteryes the bugfix goes in the gui15:30
gary_posterbut meanwhile the charm is broken15:30
gary_posterfrankban, bcsaller starting at X:31 :-)15:30
frankbangary_poster: so you are right, better to revert the changes. Even better if someone can confirm what I see.15:30
hazmatteknico, http://bazaar.launchpad.net/~hazmat/juju/rapi-rollup/view/head:/juju/agents/api.py#L15315:40
teknicohazmat, great, thanks15:41
Makyobcsaller, I think I figured out the resize thing.15:54
bcsallerMakyo: ha, sweet15:54
Makyobcsaller, it's a little unfortunate :|  Resize events are only fired when bound using Y.on(), which we're not doing with the module event binding because of publish.  If I bind it in initializer in mega.js how it used to be bound, it works.15:55
bcsallerMakyo: look at the change in the yui bindings in d3-comp though, it should cover that case, thats where I was getting stuck15:56
Makyobcsaller, Ah, I see it now.  Passing window as context gets the event to fire, at least, but then this is set to the YUI window node, rather than the module.  Is that the problem?15:59
bcsallerthe handler won't be able to pull the state it needs to run I think15:59
goodspudhazmat, let me know when you are free to talk about PPAs, login, etc16:01
hazmatgoodspud, ack, 5m max16:01
hazmateta16:01
goodspudcool16:01
hazmatgoodspud, juju-ui hangout16:09
Makyobcsaller, after target = Y;, add handler.context = null; and it works.16:10
MakyoNot...sure why it doesn't like the context in there.16:10
bcsallerMakyo: changing the context on a window event breaks it maybe... that is odd16:11
gary_posterteknico, would you like to have a call, or postpone to tomorrow, or...?16:16
Makyobcsaller, does that get us closer to where we need to be?16:18
bcsallerMakyo: I think, let me finish testing it 16:18
teknicogary_poster, a call today would be better, I'll get back to you in a few minutes, thanks16:19
bcsallerMakyo: yes, thanks alot, that working16:19
MakyoWhew16:19
bcsallerMakyo: the test is still failing, but I can track that down now that the event is properly handled16:20
bcsallerMakyo: got it, going to push again :)16:22
frankbangary_poster: re tls, e have another problem: we want, in the GUI to fetch data from http://jujucharms.com/16:22
bcsallergary_poster: makyo found the issue and I've pushed the updates, should I go ahead and land it now?16:24
teknicofrankban, I guess we can ask webops to enable https://jujucharms.com/ :-)16:37
teknicogary_poster, re: call, main thing would be making tests work, if you could help me there it'd be great, otherwise we can postpone16:46
gary_posterbcsaller, please land, yes16:55
gary_posterthank you16:55
gary_posterfrankban, :-( hazmat do you have control over jujucharms.com?  Can you set up a cert and give us https access?16:56
bcsallergary_poster: woot :)16:57
gary_posterbcsaller, reminder to please make the bug/card to fix the initial prod test run, or ask me to (happy to)  :-)16:58
gary_posterteknico, https://codereview.appspot.com/6977043/#msg5 has Francesco's charm debug notes.  They are very good.16:59
teknicogary_poster, yes, I read that email and saved it for future reference :-)17:00
gary_poster:-)17:00
teknicogary_poster, at first sight it did not help with my problem though, I'll have another look17:00
gary_posterteknico, pair for a few minutes?17:01
gary_posterup to you, only if you want to17:01
hazmatgary_poster, i do .. what do you need?17:01
teknicogary_poster, sure17:01
hazmatgary_poster, i'm  a little confused17:01
hazmatgary_poster, g+ for  a few minutes ;-)17:01
gary_posterhazmat, GUI over https needs to talk to jujucharms17:01
gary_posterok :-)17:01
hazmathmm17:01
hazmatit would be good to do so.. yes17:02
hazmatbut required.. not sure.. 17:02
hazmatdoes the browser toss up any thing for async ajax request 17:02
hazmatif its not https from an https page?17:02
hazmatwell not required atm is the questoin17:02
gary_posterdon't think so, in Chrome anyway.  frankban, ^^ ?17:02
hazmatif not then its more a pending task that we can toss in the kanban .. just trying to determine immediacy of action17:03
gary_posterI mean, I think it fails silently.  frankban has immediate experience17:03
hazmathmm.. i guess the websocket is a subresource that fails .. so its similiar17:03
gary_posterright17:04
hazmatbut thats still a bit different since that what's we wanted encrypted17:04
frankbanhazmat: each request to connect to jujucharms is blocked by the browser, including charm info (e.g. one of the service detail tabs).17:04
gary_posterfrankban, silently blocked, yeah? in chrome?17:04
gary_posterI see silent blocks in chrome, generally, and warnings you can bypass in FF17:05
hazmatthe issue with the websocket was its encrypted but with an non user auth'd cert17:05
hazmathmm17:05
frankbangary_poster: in chrome yes, errors can be seen only in the js console17:05
hazmatfrankban, what's the error msg in the console?17:05
frankbanhazmat: similar to this one, but pointing to jujucharms.com: http://pastebin.ubuntu.com/1452506/17:06
hazmatwe're sending cors headers but their use case is the reverse..17:06
hazmatfrankban, gotcha17:07
hazmatok so its immediate. checking cert prices17:07
frankbangary_poster: commented the code enabling TLS in the charm, last test run and then I'll land the branch17:09
gary_posterthanks frankban 17:09
teknicogary_poster, progress! I see machines running already :-)17:18
gary_posterteknico, excellent :-)17:18
gary_posterfrankban, do you have a plan for attacking the gui https bits we talked about?I might be able to push it forward after your EoD if you do.  If not, no worries17:46
frankbangary_poster: I think a plan could be to found a way to configure the YUI loader so that it never tries to download dependencies from the outside world17:48
frankbans/to found/to find/17:48
gary_posterfrankban, right.  That's what thiago tried at one point.  That's probably the thing to try first17:48
bacbcsaller i started reading your code to understand what you did and then ended up doing a review.17:53
bcsallerbac: thanks, its submitted, but I can make changes in the next branch if need be17:53
* gary_poster restarts then lunches17:53
bacbcsaller: ok, that's be good17:53
=== gary_poster is now known as gary_poster|away
goodspudHappy Yule Time Hugs everyone. Enjoy your holidays. I'm out of here.17:56
=== gary_poster|away is now known as gary_poster
gary_posterbenji, when you have a quick sec to explain/remind what the expected release pattern is, please lemme know18:45
frankbangary_poster: the branch finally landed \o/ . It seems to me that some modules (like gallery-ellipsis or gallery-timer) are not included just because they cannot be found under node_modules/yui.19:04
gary_posteryay frankban 19:04
gary_posterhm19:04
gary_posterfrankban, so they are not in the compressed file either19:05
gary_posterso perhaps we just copy them over and serve them19:05
gary_posterI'll try that :-)19:05
gary_posterif I have time19:05
gary_posterthanks and have a good evening frankban 19:06
frankbangary_poster: it seems so (they are not in all-yui.js). you too19:07
benjigary_poster: I am back from lunch and available at any time.19:19
gary_posterthanks benji.  ony 9 min till next call, so maybe later :-)19:21
benjigary_poster: k19:21
gary_posterbcsaller, ping20:26
bcsallergary_poster: hey, whats up?20:27
gary_posterhey bcsaller.  I am doing exploration on the https issues and messing around with things.  modules-debug.js gives this warning, which I vaguely recall:20:28
gary_poster/ The "requires" property should not be used here because the javascript20:28
gary_poster/ minimizer will not parse it.20:28
gary_posterjuju-topology requires files here, and then dupes those in its own definition20:29
gary_posterIs it OK with you to rip out the ones in modules-debug, since AFAICT they are a dupe, and can't be used?20:29
gary_poster(when compressed)20:29
bcsallergary_poster: yeah20:31
gary_postercool thanks bcsaller 20:31
bcsallerthank you20:31
* hazmat gary_poster fingers crossed23:52
hazmatwhhops23:52
hazmathappy holidays folks23:52

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