/srv/irclogs.ubuntu.com/2013/08/27/#juju-gui.txt

rick_hhatch: kinda, fish tank cleaning00:19
rick_hwhile we were chatting my son dumped all the fish food into the tank00:19
rick_hemergency cleaning in progress :/00:19
rick_hhatch: what solution is this? 00:19
hatchlol, well it's pretty trivial tbh heh https://gist.github.com/hatched/2dd9628d46be512cc75300:21
hatch^ rick_h but it comes with an odd bug, if I click 'browse' after viewing the inspector it shows the sidebar opened with only the categories....not the fullscreen00:21
rick_hlooking00:24
rick_hyae, not sure what it's doing. Will have to debug the viewState and see what's getting set00:26
rick_hthis seems strange imo00:26
rick_hwe don't need the 67-70 stuff I thought, just left over from debuggin?00:26
hatchhad to add it back in because it wasn't being set properly00:27
rick_hah, if things are minimized, the old viewmode is never set to _XXXX 00:27
rick_hmakes sense00:27
rick_hyea, die hidden stuff die00:27
hatchI'm glad you understand it because I sure don't heh00:27
hatchso can i push this up and then email it you to to take over in the am?00:28
rick_hwell, if we're updating the viewState to sidebar, then the next request we try to remove _sidebar.destroy()00:28
rick_hbut we never set this._sidebar because it was hidden00:28
rick_hhatch: sure thing, I think I understand the issues enough to get it going. 00:28
hatchok great - I really want to use every available hour to try and get this release out tomorrow and there are still the two other bugs00:29
hatchthanks - I'll push and email to you00:29
rick_hhatch: rgr00:29
=== schwuk_away is now known as schwuk
=== schwuk is now known as schwuk_away
rick_hluca__: morning, so with your email, the goal is to setup an collapsable accordian effect then? With the three headings always visible and the content of them scrollable?11:09
luca__morning rick_h 11:13
luca__sort of, Jeff knows what we are after, are you working on it now or can you wait to talk to him?11:14
luca__rick_h: because he did it wrong to start with and Gary got a bit annoyed over the wasted effort11:14
rick_hluca__: we can wait. not a problem. I'll be getting back to it soon (late today/tomorrow?) and wanted to clarify what the goals were. 11:14
rick_hluca__: ah, this is the sticky header business?11:14
luca__rick_h: yes :)11:15
rick_hluca__: nvm, I got it then. I took a stab at it at one point so I gotcha. 11:15
frankbanhey hazmat: when you have time, could you please take a look at https://code.launchpad.net/~frankban/juju-deployer/guiserver-changes/+merge/182369 ? 12:07
rick_hjujugui small review please for when people get in and need something to read with their coffee https://codereview.appspot.com/13276043/12:13
gary_posterrick_h, did you resolve with Jeff or Ben why it used to be a databinding?  The ony value I can see to the databinding is if the charm is updated and the icon changes, I guess?12:15
rick_hgary_poster: the issue before was that there wasn't the store doing icon stuff and I think it started out hoping we could generate the paths/logic in there12:17
rick_hgary_poster: but no, I didn't check on why databinding. I remeber talking with bcsaller when he added it and it was a 'show something' because the store integration wasn't ready yet12:17
gary_posterok cool.  if it is not clear how to do it with databinding then the win seems relatively small and I'm fine with the nice-n-simple approach you have here12:18
rick_hgary_poster: yea, I was going to do data binding thinking it was a case of "if the user opens another charm we can reuse the rendered html and update the data displayed with data binding"12:19
rick_hhowever, the *only* attribute data bound was the icon12:19
rick_hso I think it was a case of 'let's just hack this in' gone wild :)12:19
gary_posterlol12:19
gary_posteroh so that code is *only* for the ghost?  the post-deploy is different code?12:20
rick_hgary_poster: no, I think that it's shared. However, the post-deploy can generate the icon the same way12:21
rick_hor I just broke the normal way...so take that review request back :/12:22
gary_posterrick_h, I dunno, I could go either way.  I assumed it was shared when I said "if it is not clear how to do it with databinding then the win seems relatively small and I'm fine with the nice-n-simple approach you have here."  OTOH if you are fairly convinced of the win for the post-deploy scenario, I'll leave you to it12:26
rick_hgary_poster: yea, so I'm not sure how it fits atm. Basically I've found that from the store the charm model has a charmId field to generate the icon. 12:26
rick_hfrom the already deployed charm that attribute isn't in the model there. 12:26
rick_hhowever, charmUrl is in both, and that works to generate the right icon url12:27
rick_hgary_poster: so working on adding a test to both cases so that if this breaks it shows up12:27
rick_hgary_poster: e.g. I broke the icon in the non-ghost case and so adding a test to say it's fixed. 12:27
gary_posterrick_h, ok cool.  I'll leave the branch alone for now then.12:28
=== schwuk_away is now known as schwuk
rick_hgary_poster: got it figured out. Missed the service vs charm model case handling in there. Updated with an extra test to catch. 12:55
gary_postercool rick_h, looking again12:55
luca__Morning gary_poster, welcome back13:06
abentleysinzui: Could we do a catch-up chat this morning?13:06
abentleybac: in r362, it looks like you've changed use_context so that __exit__ is run even if __enter__ fails.  Is that intentional?13:08
gary_posterhey luca__, thank you :-)13:12
bacabentley: it was my intent to register the cleanup early but didn't take into consideration what happens if __enter__ failed.13:14
abentleybac: Why did you want to do that?13:14
bacabentley: i was working on the problem that adeuring now has with spurious test failures and it looked like the test index was not properly being cleaned up13:15
sinzuiabentley, sure. I can telling about how charmworld tip is incompatible with 2 migrations and It is time to upgrade pyelasticsearch13:16
abentleybac: Ah.  I think we should change it back, though.  It was registering the cleanup as soon as it was safe to do so.13:17
bacabentley: i can do that in my current branch13:17
abentleysinzui: Now?13:17
sinzuiabentley, sure13:18
gary_posterhey luca__ I just got to your "Juju business assumptions" email.  I'm guessing I should not bother to reply since the deadline is long past now? :-)13:58
luca__gary_poster: Heya, no need now, but thanks anyway :D13:58
gary_posterluca__, cool :-)13:58
sinzuibac, benji, rick_h, The charmworld deployment failed because r363+ is incompatible with two migrations in earlier revisions13:59
rick_hhatch: when you're around have a fix for the issue from yesterday. http://paste.mitechie.com/show/1007/ is the diff and lp:~rharding/juju-gui/browser-fix-1217060 is the branch. Adding some tests and will have it up for review later. 13:59
benjidarn13:59
rick_hhatch: then I'll be taking a strong bath for the evils committed13:59
rick_hsinzui: :(13:59
sinzuiI will ask for 3 deployments today to deploy each migration at the point it has code.13:59
sinzuithat was compatible14:00
rick_hmigrations are fun! :)14:00
sinzuiwell maybe to because I think one migration was a correction for m1614:00
benjiwe should think about how to avoid this problem in the future14:00
rick_hsinzui: time to rollup and redeploy on -core? :)14:00
sinzuiNot today14:01
gary_posterright, benji.  one approach is that we should treat a migration file as a migration in production, and devs have to deal with figuring out how to test re-migrations14:01
gary_posterthat seems simplest to me, and every other idea I have is a more complex approach for questionable benefit14:02
benjigary_poster: I'm trying to remember how we ended up dealing with migrations (or "generations" back then) in the Z4M project.  We ended up in a place that worked really well, but I'll have to remember the details.14:02
gary_posterbenji, IIRC it was the open source zope generations.  the trick was that ZODB would remember the generation it had, and then run through *all* generations14:03
gary_posterfrom last run to current14:03
gary_posterbut that is still potentially riskier in a DB/env not as flexible as ZODB14:04
gary_posterreasonable but not sufficient, I'd suggest, IOW14:04
benjithat might be how the mechanism worked, but we came up with a set of procedures that actually avoided most of the mechanism but still worked well; it'll take a while for me to remember14:04
gary_posterok14:04
hatchrick_h: ahh you went with the controls binding approach14:06
rick_hhatch: yea, I think that contains the 'evil' the most14:06
rick_hhatch: I didn't follow where you were headed with the removing of the whole check entirely and such tbh14:06
hatchohh, all I was doing was minimizing the sidebar when the user opened the full inspector14:07
rick_hhatch: yea, but then you still have the button to un-minimize visible and such14:07
hatchso the check was no longer needed becasue the browser was still on top of the inspector14:07
rick_hhatch: so there's UX conflict14:07
gary_posterbenji could you add Friday card about topic pls?14:07
hatchshhhh14:08
hatchthey wouldn't know14:08
rick_hjujugui for the later morning coffee crowd, some light reading https://codereview.appspot.com/13276043/14:08
gary_posterrick_h, ugh, I got lost14:08
rick_hgary_poster: I'm sory you get lost? 14:08
gary_postergive review to someone else rick_h 14:09
rick_hah, you mean review stuff. All good. 14:09
rick_hgary_poster: yea, I'll but the western peeps :)14:09
gary_postercool14:09
rick_hor jcsackettcif he's aroud today14:09
rick_hgrrr mobile internet lag14:09
hatchrick_h: i'll take one of the reviews14:09
rick_hhatch: ty much14:09
rick_hhatch: note this is a diff card than the one we've been going over. 14:10
rick_hjust fyi to avoid crossing the streams14:10
benjigary_poster: card added14:12
gary_posterthx14:12
hatchmorning antdillon14:36
antdillonMorning hatch14:36
bacmark uds keynote talking about cloud-ish now.  http://summit.ubuntu.com/uds-1308/meeting/21887/intro-and-keynote/14:50
hatchthx14:51
hatchhey it's the GUI!14:55
sinzuibac, benji, your revisions are now in production. I think we want a branch to remove the migrations since we know they are not all compatible with tip14:56
benjiyeah, I think so14:57
benjisinzui: I'll make a card14:57
Makyo"our less-fortunate brethren" Hah.15:01
hatchhaha yeah I laughed15:01
hatchantdillon: do you have any idea what cellular frequencies are available there?15:02
hatchmy google fu is returning conflicting details15:02
antdillonhatch, GSM 900 or 1800 15:06
antdillonhatch, http://www.ofcom.org.uk/static/archive/ra/topics/pmc/document/licencetypes/cellularinfo.htm#415:06
hatchahh nice the HTC One's in Canada have 850/900/1800/1900 GSM15:07
rick_hantdillon: is this rumor of sims in vending machines at the airport true?15:08
rick_hantdillon: it seems like a candy land of mobile connectivity15:08
antdillonrick_h, Almost, there is all way a booth selling sim cards15:09
hatchawww right on15:09
hatchguess I'll be going to get the new phone at lunch and unlocking at night haha15:09
antdillonrick_h, Or giving them away as long as you put some credit it on it15:09
rick_hantdillon: cool, yea I'll have a gsm mifi that'll need some sim love 15:10
rick_hsince I don't have a GSM phone :( 15:10
rick_hstupid verizon15:10
* hatch steals the wifi from rick_h15:10
rick_hhatch: sure thing. It's what I did before. Wifi'd my phone to the mifi and used it to get directions to dinner!15:10
antdillonI can see some wifi wars coming up15:10
rick_hhah15:11
hatchhaha nice!15:11
hatchI'm going to get the HTC One because there are awesome Android 4.3 roms for it, ANNND Ubuntu Touch port :D15:11
rick_hcool, I'm still holding out for nexus 5. I'm not due up for a couple more months. 15:11
rick_hbut then I'll be switching to t-mobile and I'll have gsm 15:11
rick_hso less of an issue with this traveling15:11
bachatch: i recall last time i was there one of the good deals on pre-paid sims was at tesco, the grocery chain15:11
hatchrick_h: yeah I'm sick of waiting, which means a new better phone will come out next week15:12
hatchlol15:12
rick_hbac: when I tested out the mifi here on ATT to make sure it worked it cost me $30 for 300mb good for 7 days. I've got low expectations. 15:12
bacbenji: would you have time to look at https://code.launchpad.net/~bac/charmworld/link-to-basket-on-lp/+merge/18241515:12
benjibac: sure15:12
rick_hbac: in copenhagen I got 1gb of data for 12months for $1315:13
hatchhahaha15:13
bacrick_h: yeah, we are ripped off and don't realize it15:13
rick_hbac: oh we realize it, but lack of options15:13
bacrick_h: in VN i could go 3 weeks with heavy data use (i get lost a lot) on a $2.50 top-off15:14
benjibac: your branch looks good15:18
bacthanks benji15:19
rick_hjujugui and hatch in particular, review #2 please for giant hacky bug. Please look away while reading. https://codereview.appspot.com/13282043/15:21
rick_hand time to head back from the coffee shop, back in a minute15:22
hatchthanks rick_h15:24
rick_hMakyo: thanks for the review. Do you have time to look at smaller/more sane other one? https://codereview.appspot.com/13276043/15:47
Makyo rick_h sure15:48
rick_hthanks Makyo, appreciate it15:49
hatchjujugui guichat in 1015:49
rick_hjujugui and the charmworld folks, might be interseted in https://pypi.python.org/pypi/nose-selecttests/0.415:53
gary_posterlooks nice.  funny that functionality is not part of nose generally, but yay for plugins I guess15:54
hatchrick_h: after your branches land can you let me know so I can re-qa the remaining two bugs in Urgent to see if those fixes fix those bugs15:56
hazmatgary_poster, nostalgic for zope.testrunner ? ;-)15:56
rick_hhatch: rgr15:56
gary_posterhazmat, exactly :-)15:56
hatchjujugui guichat in 2min15:58
gary_posterfrankban, you around for call?16:01
jcsackettjujugui: i appear to be the only one on the call...did the appointment hangout get borked, as i can't imagine that's right.16:01
gary_posterjcsackett, try guichat16:02
jcsackettgary_poster: oh, duh. of course.16:02
jcsackettthanks.16:02
gary_posternp16:02
jcsackett<--- needs more coffee16:02
bacjcsackett: i think the link in the calendar is broken16:02
jcsackettbac: yeah, it is. i don't know why i was trying to use it instead of guichat.16:03
frankbanrick_h: https://codereview.appspot.com/1274105116:11
frankbanthank you16:11
rick_hfrankban: thanks16:12
abentleyadeuring: We may want to change our pattern so that instead of creating an index and then putting the mapping, we supply the mapping at index creation time: http://www.elasticsearch.org/guide/reference/api/admin-indices-create-index/16:24
adeuringabentley: ah, very intersting. Let me try that. It would most likely fix a problem with temp_index_client()16:25
abentleyadeuring: Doing a refresh may also give us a way to wait until the index is present: http://www.elasticsearch.org/guide/reference/api/admin-indices-refresh/16:26
adeuringright16:26
hatchrick_h: the branch fixed the two outstanding urgent bugs16:28
* hatch goes back to QA'ing16:28
rick_hhatch: woot!16:28
rick_hhatch: had hoped so. 16:28
hatchmust have been the state update and some odd race condition16:28
rick_hhatch: I think it was the same thing. The viewmode controls weren't bound right in the non-browser state16:29
gary_posterInbox zero!  kinda sorta.  I have a few emails tagged I have to deal with.  But anyway, I'm taking a lunch break in celebration.16:38
rick_hwoot! only two days, email master!16:39
hatchrick_h: found another bug16:40
rick_hhatch: stop doing that16:40
hatchrick_h: haha sorry :) https://bugs.launchpad.net/juju-gui/+bug/121744916:41
_mup_Bug #1217449: Sidebar won't minimize after viewing unit list <juju-gui:Triaged> <https://launchpad.net/bugs/1217449>16:41
hatchcan you look at it now?16:41
rick_hhatch: reviewing right now, can afterwards16:41
hatchok assigning to you16:42
rick_hrgr16:42
frankbanguihelp: anyone available for reviewing a js branch? https://codereview.appspot.com/13286043 This should land asap, before further env/ws tests are added.16:42
hatchfrankban: I'll take one if no one else does (just want to finish this QA pass)16:43
frankbanhatch: thank you16:43
hatchrick_h: probably related https://bugs.launchpad.net/juju-gui/+bug/121745016:45
_mup_Bug #1217450: Browse button shows empty sidebar after clicking Juju logo <juju-gui:Triaged> <https://launchpad.net/bugs/1217450>16:45
rick_hfrankban: looking16:47
frankbanrick_h: thanks16:47
hatchfrankban: do we have tests for the Go portion of these tests which you set the default to be 'python' ?16:48
frankbanhatch: yes, I believe so, everything is more directly tested in the test_env_* files.16:52
hatchgreat16:52
frankbanhatch: re (actual, expected), is it a JS guideline?16:57
hatchfrankban: it's the syntax for the nodejs assertion library16:57
hatchhttp://nodejs.org/api/assert.html16:57
hatchfor the full api16:57
hatchit doesn't cause any issues, just creates confusing error messages when the tests fail :)16:57
frankbanhatch: interesting, I am used to expected first, and I am pretty sure I always did that in all my js branches :-) the error messages are usually confused by default, especially when you use deepEqual :-)16:59
hatchoh I know...I also think that expected should be first16:59
hatchthat's why I left the decision to switch it to you :D17:00
frankbanthanks hatch, perhaps we can clean up this stuff later with a mechanical branch17:01
hatchsure thing17:01
hatchrick_h: I'm taking off for an early lunch I believe the two tickets in urgent are related17:08
hatchbut I*17:08
rick_h hatch yea, looking at them now17:08
hatchgreat thanks a lot!17:08
=== schwuk is now known as schwuk_away
hatchbbl17:08
benjiabentley: do you have a few minutes to review my "respect charm versions in APIv3" branch? https://code.launchpad.net/~benji/charmworld/versioned-charm-api/+merge/18245017:18
abentleybenji: looking...17:40
hatchrick_h: back17:58
hatchneed any help with that stuff now?17:58
rick_hhatch: no, think I've got it. Just sanity checking and debating on testing/etc17:59
rick_hhatch: fixes both of them17:59
hatchawesome thanks17:59
* hatch picked up the HTC One17:59
hatchchose a new store this time and was in and out in 10mins17:59
rick_hhatch: very cool, jealous. You'll have to let me check in out next week ;)18:00
hatchyep, now I'm stuck in a 3 year18:00
hatchhaha18:00
rick_h3yr?! ouch!18:00
rick_hI get ancy before 2yrs18:00
hatchit's not like i'm going to go with another provider so it's just picking the option that made the phone the cheapest18:02
hatchit was $10018:02
hatchI'm pumped that I am able to disable apps now18:03
hatchbye bye Facebook18:04
hatchalthough after I get back it'll likely be getting a rom anyways18:04
rick_hhatch: I've got one issue I'm not able to see wtf atm18:07
rick_hhatch: got a second?18:07
hatchyup18:07
hatchin guichat18:08
hatchrick_h: about to test18:38
hatchsorry make hung18:38
rick_hhatch: ok, I've got a couple test failures from it working out18:38
hatchrick_h: I got it to fail but then if I let it sit, the gui portion goes away18:39
hatchlol18:39
hatchit's like a few second delay18:39
rick_h*UGH*18:39
rick_hhatch: make sure to clear cache and such18:39
rick_hI don't know...just trying to find some excuse it could be something else18:39
hatchnow I can't repro :/18:41
hatchrepro'd18:43
hatchnow how...18:43
hatchit's like there is a race condition18:44
hatchsometimes I can see a 'flash' of the inspector as the sidebar comes up18:44
rick_hhatch: k, well hack inbound. Give me a few min18:45
hatchalright thanks18:46
abentleybenji: sorry for the delay.  Was ambushed by a UDS session.18:59
benjiCRIKEY! Isn't she a beauty?!19:00
* benji watches abently wrestle with the wild UDS session.19:00
rick_hhatch: guichat?19:08
hatchthere19:08
hatchrick_h:  :(19:31
hatchdbl click > Build > Juju (...and repeat) still happens19:31
hatchrick_h: does that not have an issue for you still?19:34
rick_hhatch: looking19:34
hatchthe dbl click is on the service of course19:35
rick_hhatch: ok, something is setting renderEnvironment to false. Looking19:36
rick_hhatch: ok, so when I navigate with clearAllNameSpaces, it's hitting back into the toggleStaticViews with a url :gui;/service...19:38
rick_hthen I get the message 'waiting on service data' and that causes anothewr dispatch to go through that clears it19:39
rick_hhatch: so my navigate() command isn't 'die gui die' enough19:39
hatchlol19:40
rick_hhatch: so how would I generate a _navigate that manually set :gui:/19:40
rick_hvs :gui:/service...19:40
rick_h?19:40
hatchthere is a url method on the nsrouter19:42
hatchwhich you pass a config object to19:42
hatchand it generates a url for you19:42
hatchsee line 1307 in app.js as an example19:42
rick_hlooking19:43
abentleybenji: "Featured bundle UI" looks redundant with "User innterface to (un)set a bundle as featured".19:48
benjiabentley: it does, are those card titles?19:49
abentleybenji: Yes.19:49
abentleybenji: You're assigned to one, but not the other.19:49
benjiabentley: will you kill one of them for me?19:49
rick_hhatch: I don't get this...I navigate to a fresh url but the url in req.url in toggleStaticViews is still :gui:/service...19:49
abentleybenji: Done.19:50
benjithanks19:50
abentleysinzui: Do you know what "charmworld should serve README in same pattern as charm" means?19:50
hatchrick_h: so the url in the address bar is empty?19:51
hatchminus the hostname19:51
rick_hhatch: no, it's got :gui:/service/mongodb19:51
sinzuiabentley, not immediately19:51
hatchoh ok, so then that makes sense?19:51
sinzuiabentley, and not after a few moments of thought19:52
rick_hhatch: no, so I run this.navigate(new_url);19:52
rick_hhatch: and in the views the url is the old url19:52
abentleybac: What does your card "charmworld should serve README in same pattern as charm" mean?19:52
rick_hhatch: guichat again? sorry19:52
hatchsure19:52
sinzuiabentley, bundles need a readme, and the cw page should render it?19:53
bacabentley: better?  "(nice to have) charmworld should serve README for bundle in same manner as for charms"19:53
abentleybac: Much better.  I didn't understand that it was about bundles.19:54
bacabentley: i am going to mark it blocked on my current branch, though19:54
abentleybac: Okay.19:54
hatchbcsaller: kickin around?20:27
bcsallerhatch: yeah20:27
bcsallerwhats up?20:27
hatchmind poping into guichat?20:27
hatchwith rick and i20:28
benjiabentley: thanks for the good review; I've made all the requested changes -- I think ;)20:45
abentleybenji: looking...20:48
hatchbcsaller: rick_h gg :P21:11
rick_hhatch: heh, let me know how far you get. I can finish up/clean up in the morning. A simple test on the initState should work to make sure it's cleaning 21:12
rick_hhatch: that's really the only *change* that's needed is that stupid initState method 21:12
hatchhaha right21:12
rick_hI'm outta here...21:12
hatchI'll send you an email with as far as I get21:12
hatchhave a good night21:12
rick_hhatch: sorry to delay the release :(21:13
* hatch shakes fist21:13
hatchlol np21:13
rick_hhatch: or feel free to leave it, Probably not getting 2 reviewd and such to release today anyway21:13
rick_hI'll have it up ready to go before you start tomorrow21:13
hatchoh ok can do that too21:13
hatchI can move onto other things21:14
hatchok so I'll leave it then21:14
rick_hyea, up to you. I know I'm blocking you and don't want to, but not sure how much you'll get anyway :/21:14
rick_hcool, thanks for the help/talk throughs21:14
sinzuibac: your revisions are on staging21:56

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