/srv/irclogs.ubuntu.com/2013/09/11/#juju-gui.txt

* hatch attempting to update ubuntu vm00:22
huwshimihatch: Have you ever had a problem where IE sits on the "Connecting to the Juju environment" screen and doesn't get any further?00:29
hatchhuwshimi: yep00:29
hatchand i have NO Idea wtf it is00:29
hatchno errors, no nothing00:29
huwshimiyeah...00:29
hatchjust sits there and has a party or something00:29
hatchif I remember it doesn't even send the ws request00:30
hatchusually closing the browser down and re-opening it solves the issue00:30
hatchI get some very odd bugs that are solved by doing that00:30
hatchhaha00:30
hatchgo IE!00:30
huwshimihatch: That didn't work for me. Maybe I'll shut down the vm00:31
hatchhmm00:31
hatchyeah maybe try that00:31
hatchand clear the cache like 100x00:31
hatchsometimes it forgets files in there for fun00:31
huwshimihatch: Ah, clearing the cache looks like it fixed it00:36
hatchoh, heh woops I figured you would have already done that00:37
hatchhaha00:37
hatchodd ie issue? 1) clear cache 100x 2) restart browser & clear cache 3) restart system 4) install Ubuntu00:37
huwshimihatch: I don't think I've ever had a cache issue resulting in something like that...00:37
hatchevery once and a while I'll get a odd DOM issue which requires steps 1 through 300:38
huwshimiheh00:39
hatchI find it very odd that Apple, Dell, and HP are all releasing their new 'pro' haswell laptops in October00:42
hatchI wonder if this is a supply issue...00:43
MakyoIt is.  They're retiring a bunch of sockets for OEM.00:43
hatchohh really?00:43
MakyoYeah.  I mean, LGA775 went away a few years ago, and 1366 or whatever it is now, the late core2/i7/9 series is going away.00:44
MakyoFriend from Intel's been working on Haswell.  Sounds fun, if waaaay over my head.00:44
hatchahhh00:48
hatchhah yeah the processor stuff kind of blows my mind00:48
jcsackettso it's confirmed that it'll be october?00:48
jcsacketti've heard rumors, but seen nothing concrete.00:48
hatchjcsackett: no just rumors00:48
hatchbut Dell and HP both say October00:48
hatchso if their rumors say Oct and the others are definitely saying Oct00:49
hatchI'd put money on it :)00:49
hatchIt's going to be a prety interesting segment - will have to see the reviews for the Dell and HP to see if they even come close to the MBP01:04
hatchrick_h_: gary_poster so the issue with my juju-core was that 12.04 only has 1.12 which doesn't accept --'s infront of it's flags01:09
gary_posterhatch, gotcha.  You'll probably still want more recent, but maybe that is OK01:10
hatchyeah it's upgrading now01:11
gary_postercool01:11
hatchwe'll see if it bombs01:11
gary_postersuch optimism :-)01:11
hatchlol I know right?01:11
* hatch crosses fingers01:15
hatchwell it didn't totally bomb01:15
hatchI have a desktop...01:16
hatchsortof01:16
hatchyeah this upgrade didn't go well01:16
hatchlooks like it's an issue with parallels01:23
hatch*sigh*01:23
hatchyeah this vm is destroyed01:31
hatchI guess parallels does not do Ubuntu upgrades very well01:34
hatch*rainbow* The more you know */rainbow*01:34
gary_posterheh01:37
gary_posterhey huwshimi, looking at your branch.  haven't qa'd yet, but noticed that you removed yui3-g classes and friends from everywhere except https://codereview.appspot.com/13352047/diff/1/app/templates/service-configuration.handlebars?column_width=80 .  Why not there?01:38
gary_posterhuwshimi, qa good.  address my question and then land it, please!01:49
huwshimigary_poster: Ah, Did you happen to move those controls? I thought I had changed them but then I had a conflict in a merge (I thought at the time they were moved).01:53
gary_posterhuwshimi, yes, I did01:53
huwshimigary_poster: Ah, I thought I was going loopy :)01:54
huwshimiFixing now01:54
gary_posterhuwshimi, sorry. :-) I was trying to keep the structure similar between the constraints and config01:54
huwshimigary_poster: It's all good :)01:54
gary_posterthanks :-)01:54
gary_posterhuwshimi, you saw manage.jujucharms.com has your updates?02:06
huwshimigary_poster: Ooh! Must have happened overnight!02:06
gary_posterhuwshimi, yup :-)02:07
gary_posterwe're the magic elves for your side of the world, and vice versa :-)02:07
gary_posterKapil complemented it02:08
huwshimi:)02:08
=== rogpeppe1 is now known as rogpeppe
=== schwuk_away is now known as schwuk
gary_posterThanks for replying to Peter, bac11:00
gary_posterhm, bac is not hre11:01
rick_h_he's hiding11:01
gary_poster:-)11:01
hatchmorning11:55
hatchday 2 @ 6am11:55
gary_posterheh11:57
ahasenackgary_poster: hi, quick question, where is the export button in the new gui? I'm on http://comingsoon.jujucharms.com/11:58
ahasenackgary_poster: or is it not there? Export as a "bundle" to use with juju-deployer11:58
gary_posterahasenack, functionality currently is only exposed as hotkey.  button coming later.  shift-d exports. (shift-/ , or '?', will show hotkeys)11:59
ahasenackgary_poster: ah, thanks11:59
gary_posterwelcome12:00
* bac is saucy-fied12:27
* bac and wonders why juju is installed in such a dumb place12:28
hatchI attempted to do that last night and borked my vm12:39
hatchI blame gary_poster - he peer pressured me to do it12:39
gary_poster:-P :-)12:39
gary_posterhey bac.  thanks for replying to Peter12:40
bachatch: but you did take a snapshot?12:40
bachatch: mine went without a hitch12:40
hatchbac: nope I didn't have a spare 20gigs :/12:41
hatchpoor decision on my part12:41
bacoh12:41
hatchthis weekend I will set up a new one and update our hacking docs12:42
hatchas I'm sure they are horribly out of date12:42
bachatch: that's a mighty positive spin on it.  :)12:42
hatchhaha - well there has been a ton of changes in the environments and I haven't seen any changes coming through on the hacking docs12:43
baci am a bit underwhelmed with the saucy background.  i force myself to use the release-branded one so i can easily tell my vms apart but this one is terribly boring.  i'm going to miss that little ringtail.12:44
gary_posterhey hatch, does this mean you can't do reviews? :-) If not, would appreciate a review of https://codereview.appspot.com/13373052/12:47
hatchgary_poster: I can, I am just working from my laptop so things are a little slower to build12:49
hatchreviewing12:49
hatcheven with all the peer pressure I wouldn't have done that upgrade without a viable backup to work from :D12:49
gary_posterack, thanks hatch.  I have a lot of small reviews coming down the pike today.  trying to make small targeted branches that are easier to review, but that means I'll be bothering people a lot.12:49
gary_posterI mean, more than usual ;-)12:49
hatchhaha12:49
hatchno problem - I just finished the changes ben wanted me to make to my branch so that should be landing soon after a big qa again12:50
hatchnice improvements to the branch12:51
gary_posterawesome12:51
hatchqa'ing in IE now13:04
hatchgary_poster: so your branch is lgtm but conflict ux is broken in IE on trunk13:11
gary_posterhatch, before and after my branch?13:11
hatchbefore and after yes13:11
hatchthe little dropdown never shows up13:12
gary_posterhatch ack.  Do you object to me landing as is?  I have a train of branches I'm trying to...uh, bring to the, uh...station.13:12
hatchlol13:13
hatchnope I lgtm'd13:13
hatchbut I'll create a high priority card for the ie fix13:13
gary_posterok thanks hatch, perfect13:13
hatchdone and done13:15
hatchlol I just got an email from Air Canada13:19
hatch""Double your Aeroplan Miles: fly to Beijing, Shanghai, and Hong Kong!""13:19
hatchOH OK NO BIG DEAL!!!!13:19
hatchlol13:19
rick_h_hatch: got a sec? I had to change the id of the checkbox inputs and now the slider doesn't work. I'm looking at the css and not seeing how the id comes into play13:25
hatchsure, couple minutes, just need to get this branch re-proposed13:27
rick_h_hatch: rgr13:27
hatchok ready13:31
hatchI'm pretty confident that the reason phantom always crashes is because we are running too much stuff to the console13:37
hatchso if we switch to a different output and somehow stop it from showing all of the 'GET' requests13:37
hatchit should stabilize it13:37
rick_h_I've not had phantom crash13:38
rick_h_what does it crash during? test-debug or something?13:39
hatchyeah - but if I run it in the background it works 100% of the time13:39
rick_h_ah, maybe. I've not had it happen :/13:39
hatchso that's my theory13:39
hatchI remember jc was also having a similar problem13:39
gary_posterjujugui, could someone please review https://codereview.appspot.com/13348056 ?  It is another juju-gui databinding fix branch, fairly small (only one reviewer needed), with some line-by-line comments..13:45
bacgary_poster: i'm talking to peter about GA.  it looks like they have a new key they'd like us to use13:45
benjigary_poster: I'll take a look13:46
bacgary_poster: i think it might be a good time to make the GA key configurable in the charm configs.  we could separate jujucharms.com from other deployments13:46
gary_posterbac, yeah I saw in some email or other.  As long as we have access I guess it is fine.  Yes, I was thinking the same thing: sounds perfect.  you have the bandwidth for that?13:46
gary_posterthank you benji13:46
bacgary_poster: yes, i have not yet started a new task so now would be perfect13:47
gary_posterok cool thanks bac13:47
bacgary_poster: i think we should default to the old GA key but on deployment to jujucharms.com we should switch to the new one.  that's how i'll approach it.13:51
gary_posterbac, +113:51
gary_posterwas thinking same13:51
benjigary_poster: done13:53
gary_posterbenji, thanks.  yeah, I will elaborate on those docs before landing.  I was figuring it out as I went. :-)13:54
benji:)13:56
sinzuiabentley, adeuring maybe we want to tag charm-tools bugs with "charmworld" I think of charmworld as a stakeholder.14:05
adeuringsinzui: +114:05
abentleysinzui: +!14:05
abentley+!14:06
abentley+114:06
rick_h_ah crap, note to self. disabled checkboxes do not fire change events...this is important in writing tests. :/14:06
jcastrogary_poster: I wasn't expecting the new managed site to go up so quickly!14:06
sinzuiadeuring, abentley or charmbrowser, which is the same used by juju-gui14:06
gary_posterjcastro, but a good thing, yeah?14:06
jcastrodude absolutely!14:06
gary_postercool :-)14:07
jcastrohey so, are we going to move the redirects to that site? for example https://jujucharms.com/precise/wordpress/14:07
abentleysinzui: potato potahto.14:07
sinzuianyone working on the charmworld title metadata problem?14:07
sinzuiI have it is good authority that it will now be spelt potatoe14:07
jcastrogary_poster: also are you the guy to ping on manage. or is that ale/luca?14:09
gary_posterjcastro, me for now at least14:10
jcastrook so so far I think it's just the redirects that are weird14:10
jcastrolike go here: http://manage.jujucharms.com/charms/precise/appflower14:10
jcastroand then drop the "manage" and then you'll see what happens14:11
rick_h_ugh, bug there. -HEAD-1 oops14:12
jcastrowell, are we even supposed to be sending people to the GUI now?14:12
jcastroor the static pages? 14:12
jcastroI thought static pages?14:12
gary_posterjcastro, first, that doesn't sound like a manage.jujucharms.com yeah?  You are talking about jujucharms.com redirects?14:15
jcastroyeah14:16
hatchjujugui I'm still looking for an in depth qa of https://codereview.appspot.com/13373050/14:16
jcastrook so basically, when I'm telling people "hey man check out our wordpress charm" we're sending them to manage. right?14:16
jcastronot the gui?14:16
gary_postersecond, https://jujucharms.com/precise/appflower works great now fwiw and afaict.14:17
sinzuiadeuring, I tagged several charm-tools bugs with charmbrowser.14:17
jcastroright, but my question is, which is the "canonical" page for a charm on the web?14:17
gary_postershort term, the urls in m.j.c are supposed to eventually match the j.c ones (so you don't have to include "charms")14:17
adeuringsinzui: ack14:18
jcastrogary_poster: ie do I send people links to jujucharms.com or manage.j.u.c ?14:18
gary_posterjcastro, short term, I think we want https://jujucharms.com/precise/appflower to be canonical.  Soon, from orange squad IIUC (but this is up to mramm), that will proxy m.j.c if your browser can't cut it.  long term the plan is to eliminate both of these and have everything live in one place, currently proposed as juju.ubuntu.com.  That's not till post SFO though14:19
jcastrook gotcha!14:20
hatchjcastro: can people who work for Canonical enter the charm championship?14:25
rick_h_hatch: looking at it14:28
jcastrohatch: nope, sorry14:29
rick_h_jujugui looking for two reviews + qa https://codereview.appspot.com/13516044/ 14:30
rick_h_doh, guess I better resolve conflicts first14:30
hatchjcastro: booo!14:30
gary_posterheh14:30
hatchwhat if I got my mom to submit it?14:31
hatch:P14:31
hatchjk14:32
hatchrick_h_: reviewing14:34
rick_h_hatch: down to the last of 4 conflicts atm14:34
rick_h_so watch out14:34
hatchohh ok I haven't gotten that far14:34
hatchwill wait14:34
hatchhow did you propose with conflicts in the branch?14:35
hatchor did you just forget to merge trunk14:35
hatchwait...so did I14:35
rick_h_hatch: I didn't merge trunk this morning14:35
hatchheh oops14:35
rick_h_and you all were busy yesterday :P14:36
hatchhaha yeah that diff was huge14:37
rick_h_re-lboxing14:37
hatchmore haswell in Lenovo laptops http://www.engadget.com/2013/09/11/lenovo-thinkpad-business-notebooks-haswell14:42
hatchand in untypical Lenovo fashion they are actually going to be available within a few months of the release date14:42
hatchlol14:42
hatchunfortunately comments don't carry forward from the previous revisions so rick_h_ all of your comments are gone :(14:46
rick_h_oh son...14:46
rick_h_well, they're in the comment history and the links still work14:47
rick_h_will have to go with it14:47
rick_h_I tried reviewers!14:47
hatchyeah it's just unfortunate14:47
abentleysinzui, adeuring, jcsackett: I've updated canonistack-sshuttle so that it only needs an environments.yaml with correct auth data.14:55
sinzuisweet.14:55
hatchrick_h_: I was under the impression that the conflict ux reacted to elements already in the DOM15:03
hatchso that the 'skip conflict' thing wouldn't be necessary15:03
rick_h_hatch: no, there's two things of conflict. One is the conflict-pending, another is conflict itself15:04
rick_h_one shows the ! 15:04
rick_h_the other shows the chooser15:04
rick_h_the chooser is html already in the dom hidden by default15:04
rick_h_the ! is added/removed as required15:04
hatchahh ok - so what handles the ! stuff?15:04
rick_h_for other inputs the ! is a background image15:04
rick_h_for checkboxes it's a custom div15:04
rick_h_hatch: so it's part of the change/conflict methods in conflict mixin15:05
rick_h_hatch: that was updated last week to support checkboxes, this just tweaks them15:05
rick_h_to be in the right place wit the correct smaller image now15:05
hatchohh ok so it's not part of databinding it's handled by the mixin called from the databinding15:05
rick_h_hatch: rgr15:05
rick_h_hatch: is there a feature flag for the relations stuff?15:09
hatchnope it just works15:09
hatchor should ;)15:09
rick_h_hatch: I don't see the tab 15:09
rick_h_oh duh...ignore me15:09
hatch*phew*15:10
hatchI want that darn branch to land already15:10
rick_h_hatch: k, well review feedback is in. I'm doing qa work now15:10
rick_h_hmm, make clean-all to the resuce hopefully15:11
hatchafter all the bzr colo issues you have I'm pretty glad I don't do that :)15:16
rick_h_hatch: can we toss an initial "no current relations" or something in the tab? I thought it was broken at first15:16
rick_h_hatch: heh, well turns out that the tab is just empty by default and I thought it was broken15:16
rick_h_so the make clean-all wasn't necessary15:16
hatchYeah I have no objection to that but we don't have UX for it so I'd like to add that in the follow-up ux branch15:16
hatchin fact, I think it's necessary :)15:16
rick_h_yea, I understand now that I've gotten it to work. Threw me off at first. It's visible by default by nothing that says what it is. Not even a "Relations" heading or something 15:18
rick_h_I think if it lands it needs to say 'something'15:18
hatchyeah actually the design is clearly wrong - it should have some indication as to what that icon means15:21
hatchI'll add a 'Relations' heading by default15:22
rick_h_hatch: thanks, would help. Qa'ing in Go now15:23
* hatch crosses fingers15:25
hatchhopes his time reading Go source and writing the go simulator was done properly15:26
rick_h_hatch: so there's a 'always on' relation for mysql. interface mysql-ha, peer relation15:34
rick_h_hatch: so that has no heading since it's not related to another service15:34
hatchhmm15:34
hatchlemme see if I can repro in the simulator15:34
rick_h_hatch: rgr, I didn't see it in the normal sandbox qa15:34
hatchhmm nope15:35
hatchso why the difference I wonder...15:35
hatchand so it just shows the relation data but no Header?15:35
rick_h_hatch: right, I can screenshare if you want to see what I see15:36
rick_h_hmm, now mssql is in error state and I can't resolve/retry my way out of it :/15:36
hatchthat's not an issue of the branch though right?15:37
rick_h_no, just stuck now. bah. 15:37
hatch:/ darn15:37
rick_h_bah, have to destroy the env and start over15:38
hatchwhen you can...could you open up viewlets/service-relations.js with this mysql and dump the `relations` object on line 40? I think I can just do a check for the service and if 'peer' then update with the current service name or something15:38
hatchoh darn....sorry :)15:38
rick_h_I'll see what I can do, but we don't have a good way to run debug from the charm atm15:39
rick_h_so opening the file won't do me much good15:39
hatchohhh right damn15:39
hatchhmm15:39
rick_h_I can try to app.db.services()...?15:39
hatchohh right ok umm15:39
hatchone sec I'll craft a line15:39
hatchY.juju.views.utils.getRelationDataForService(app.db, app.db.services.item(?????));15:41
hatch^ rick_h_ where the ???'s are you'll need to enter the appropriate index #15:41
hatchlikely 0 if you only have mysql out there15:41
rick_h_hatch: 1 because of the gui service itself15:42
hatchoh right15:42
rick_h_hatch: will try it out once I get the gui back up and pointed back at your branch, takes a few min15:42
hatchyeah no rush15:42
* gary_poster observes hatch and rick_h_ playing dueling banjos with their branch reviews :-)15:43
hatchlol15:43
hatchrick_h_: you added comments to two different patch sets lol15:44
hatchI was like 'where the heck are these other comments?'15:44
rick_h_hatch: I give up...15:44
rick_h_hatch: I just went in and hit 'start review' then went back into individual files for follow up. e.g. didn't make comments on missing tests until I got done and saw the tests didn't apepar15:44
rick_h_appear...man is today Monday or what?!15:45
hatchcould be!15:45
gary_posterjujugui call in 1015:50
rick_h_hatch: let's hang after the call for the qa stuff. That line won't work and i'm trying to find another way to get that for you15:54
rick_h_hatch: but mysql won't deploy correctly anyway so I wonder if there's a better qa example I can run15:55
rick_h_and I wonder if there's a bug in this resolve/retry stuff15:55
hatchohh - umm you could use anything really15:55
hatchI was using mysql because I could get a list of relations15:56
hatchso what else can have multiple relations...hmm15:56
rick_h_yea, will check out others. I like mysql because of this special relation that's showing15:56
hatchyeah that's a good find - I'm curious as to why it doesn't show up in the simulator15:58
hatchwe should find THAT out first15:58
gary_posterjujugui call in 215:58
hatchbcsaller: any idea ^ ?15:58
hatchI just realized that I pointed you to a 20min convo...15:59
hatchignore it, I'll explain after the call :)15:59
bcsallerahh, ok15:59
hatchumm i think guichat is broken15:59
rick_h_man, I can't deploy elasticsearch15:59
hatchThis party is over...15:59
rick_h_hatch: yea, got the same thing15:59
bcsallerfun while it lasted15:59
benjigary_poster: uh oh: "This party is over..."16:00
hatchdarn so you can't keep links anymore :(16:00
rick_h_call https://plus.google.com/hangouts/_/c989444bba77a138fe2d2ba091887b7dea5739a1 for now 16:00
gary_posterhuh16:01
BradCrittendener?16:01
gary_posterBradCrittenden, see link from rick_h_ 16:01
=== BradCrittenden is now known as bac
=== marcoceppi_ is now known as marcoceppi
bacso is tinyurl.com/jujugui dead?  if so, can we relink it?16:15
gary_posterbac, it is dead, and no, I don't think tinyurl.com lets you repoint (unless you pay money, maybe?  some systems are like that).  doublechecking...16:16
bacdoesn't matter much.  we just need a persistent hangout url we can bookmark16:17
gary_posterhey jujugui, anybody have decent RoR experience?16:19
bcsallernot much more than HelloWorld16:19
hatchsame16:21
hatchI know a guy...16:21
bcsallerI know django, same thing, different language 16:21
rick_h_hah16:22
hatchlol16:22
gary_posterjujugui, guichat is dead, long live guichat: http://tiny.cc/guichat16:27
gary_posterI can edit that if Google kicks us off again in the future16:27
hatch:'(16:27
gary_postertinyurl does not allow you to edit16:28
hatchrick_h_: did you create the bug or would you like me to? (re peer relations in fakebackend)16:29
rick_h_hatch: I did not create a bug16:30
gary_posterjujugui, another small databinding branch ready for review: https://codereview.appspot.com/13663043 .  I would appreciate someone running tests in IE 10; my test might suffer from the problem hatch reported there.16:32
* gary_poster goes for some lunch16:32
hatchrick_h_: bcsaller both of you mentioned the 'hacks' to get around the tests which were trying to be 'unit' tests but were really integration tests16:56
hatchso...should I make the tests true integration tests?16:56
hatchthere are quite a few16:57
hatchwill probably push this branch out at-least another day or two16:57
rick_h_hatch: yea, how many are 'quite a few'? I was hoping it'd be a matter of updating setup/teardown16:57
hatchunfortunately no :(16:57
rick_h_hatch: though there were several different places it looked like16:57
hatchI agree the tests should be fixed16:57
rick_h_hatch: so yea, I think fixing the tests is important. The thing that is the silent failure potential otherwise. 16:58
hatchhow about....I tag each 'if bloick' which was a hack so that a quick follow-up could be done to fix it properly16:58
hatchyeah that's a good point16:58
rick_h_hatch: if they console.error or something I'd be more likely to say short term ok16:58
rick_h_I just get nervous that there's a silent "it worked" and leads to a really tough bug to find later on16:58
bcsallerconsole.error makes more sense than adding a day to the branch16:58
hatchok that's good - so console.error and a flag to make it easy to track down16:58
rick_h_hatch: I can get on board with that16:59
hatchrick_h_: yup that's very true - it would be very hard to find haha16:59
hatchrick_h_:  I have no idea how the heck you commented on two different patch sets lol17:19
rick_h_hatch: I told you how 17:19
hatchright but that's funny17:20
rick_h_Yea, lesson learned there. Some times the linear process of reitveld makes for fun17:21
hatchIts too bad it doesn't appear to be under development any longer17:21
hatchit's so close to being awesome17:21
rick_h_we could try out reviewboard :P17:22
rick_h_very active dev there /me is a fan17:22
hatchnever heard of it17:22
* hatch checks17:22
hatchhmm does look pretty nicely featured17:22
hatchdoes it work with bzr diffs?17:23
hatchI can't seem to find a y/n17:23
rick_h_think so, I have not tried it. 17:23
* hatch adds to list of "things to never get to because there isnt' enough time even though he wants to look at"17:24
gary_posterthank you very much rick_h_ 17:24
rick_h_gary_poster: np17:32
hatchgary_poster: in hangout, whenever you're ready17:59
* benji is begining to enjoy sending multi-dozen page faxes to the bank.18:28
* benji looks into Stockholm syndrome treatment options.18:28
hatchbenji: buying a house?18:33
benjihatch: trying to sell one (short sale)18:33
hatchoh interesting - I didn't know what a short sale was18:35
* hatch is curious how the topology relation test file is only 116 lines when the topology relation file is 1137 :)18:52
rick_h_http://www.seejohncode.com/2012/03/13/setting-up-mocha-jscoverage/18:55
hatchrick_h_: I think we have an architectural problem with our tests that I"m not entirely sure how to solve....but I think a 'mock' factory is a good start18:56
rick_h_hatch: example?18:56
rick_h_e.g. I didn't 100% follow that last statement18:56
rick_h_or not e.g., translation something18:57
hatchso say I want to test that I can remove a relation18:57
rick_h_rgr18:57
hatchor that a method removes a relation from the db18:57
rick_h_doorbell, sec18:57
hatchso in my test I would push something into the db, then execute that method and see if it's gone18:57
hatchbut what's happening now is that to do this properly I need two services and properly constructed relation18:58
hatchso I'd need to mock that in every test suite that I need this in18:58
hatchwhat I should be able to do is something like18:58
hatchservices = mymocks.createRealService();18:58
hatchmymocks.createRealRelation(services)18:58
hatchbecause in so many places in our tests we are creating local mocks with only the required data18:59
hatchso the mocks can get stale, or if something changes, it needs to be changed everywhere18:59
rick_h_right, so in python we'd have factory helpers makeService(), makeCharm() and maybe helpers like makeRelation(service, service2) etc19:00
rick_h_and then those factory helpers would have cfg options to set/adjust things as required so that you can put in the data you care about, and fake the rest19:01
hatchexactly19:01
hatchright now I'm looking at creating a test that removes a relation and it's just a *groan* activity because I have to make all these mocks19:02
rick_h_http://bazaar.launchpad.net/~juju-jitsu/charmworld/trunk/view/head:/charmworld/testing/factory.py in charmworld19:02
rick_h_hatch: so yea, so two parts. 1) is the code setup in a way in which you can hit the method directly and if not, can it be. e.g. a single function call that takes the two services or something19:02
rick_h_hatch: and then 2) a mock helper that we stick into a test utils module that can help along the lines of a factory19:03
hatchAND we also have a lot of functions which are......multipurpose..... :)19:04
hatchmakes them hard to test19:04
rick_h_hatch: but this is exactly why I want to keep events/viewlets self contained. Just objects I can init/call/test. I don't want to have to keep creating mock environments, stores, dbs, etc. 19:04
rick_h_too often it is a giant pain19:04
rick_h_hatch: yea, so some of it is that whole design with tests from the start so that you end up breaking things up nicer to be able to test19:04
rick_h_and some of it is adding in the helpers to make it easier as well19:04
rick_h_it'll always be easier to test a single function I can pass in a couple of custom objects to, than to have to create those objects, stick them in a db, update some db state, and then call some other object.doSomething()19:05
hatchwell you need integration tests19:05
rick_h_and doSomething looks in the db, vs just doSomethign(obj1, obj2) and the call is this.doSomething(db.find(1), db.find(2))19:06
hatchthere is a thought in testing that you should never unit test19:06
rick_h_hatch: yea, this isn't against any of that19:06
hatcha line of thinking I mean19:06
hatchI don't think I agree you can be so black& white about it19:06
hatchcan't*19:06
rick_h_yea, it's not black/white. I tend to go larger than a unit, but I do find that there's smaller blocks than we typically use that would work. 19:08
rick_h_the widgets/view stuff in the browser I'm a big fan of. Test widget will fire event X with payload Y when you do Z. 19:08
rick_h_then go test the view, given an event payload of Z, verify that the view will do A and B19:08
rick_h_trying to instantiate the widget, the view, the model, the store, etc...in one test is a pita and you'll end up skipping over giant swaths of cases19:09
rick_h_hatch: if you want to pair on something specific I'd love look at it and see if I can suggest anything. Maybe we can start a factory helper. 19:10
rick_h_hatch: if not, cool. I agree with your general concensus/idea19:10
hatchwe need a sprint19:10
hatchwhere we tackle modules and get 100% test coverage and proper splitting upping of the functions :)19:10
rick_h_yea, I'd love to get coverage working. Maybe I'll try to do that as a slack thing. At least to get some ideas of where we're at. 19:12
rick_h_that js coverage module seems kind of a pain as it wants to be setup per directory or something, will have to try it out19:12
hatchyou'llprobably cry19:12
hatch;p;19:12
rick_h_oh, I already do :P19:12
rick_h_but I do the same things. All too often I only add a test for the success case and I've got to stop doing that.19:13
hatchbcsaller: still around?20:39
bcsalleryeah20:39
bcsallerwhats up?20:39
hatchjust proposing that branch right now so would love a follow-up20:39
hatchhopefully to get the darn thing landed20:40
hatchall of the code paths are 'tested' but there is no way to know that....we really need to develop a better testing strategy20:40
hatchI call for a few remote beerz discussion about it some time :)20:41
hatchor maybe in....whatever city the sprint is in lol20:41
hatchok it's done proposing20:43
hatch^ bcsaller20:52
bcsallerhatch: looking for the link now20:52
hatchwoops sorry https://codereview.appspot.com/13373050/20:52
hatchthought it would have emailed20:52
hatchbcsaller: thanks for the review and sorry about missing that - I meant to move that into the service21:10
bcsallerhatch: np21:13
hatchsubmittin!21:18
hatchalthough I have two followups so I'm not really out of the woodwork yet21:18
hatchhah21:18
rick_h_looking now hatch 21:18
hatcha new branch will immediately be created for the unit aggregation21:20
hatchwhich will add some normalization and more tests21:20
hatchwell immediately being in the AM because it's past EOD and I have a bunch of stuff to do tonight21:20
hatch:)21:20
hatchI'm really going to work hard on doing TDD and small branches21:21
hatchneed to land things after a dayish21:21
hatchthese long runners just kill forward momentum21:21
rick_h_don't go too crazy. I find pure TDD tends to end up with bad apis, but definitely smaller/jumps21:21
hatchdid you read my TDD blog post?21:21
hatchI like that approach21:22
rick_h_hatch: yea, that was a while back I think. 21:22
hatchyeah21:22
hatchI find it's a more common sense approach to tdd21:22
rick_h_coolio21:22
hatchI'm also downloading 13.04 so I can hopefully start working on my desktop again soon hah21:23
hatchlaptop is great but man I need moa powa!21:24
rick_h_yea, I can understand that21:24
rick_h_hatch: so I'm looking this over and cool. I see the new test for the model initializer, the one that hits the bound event *add. 21:24
rick_h_I don't see anything for the large swath of process_delta or large, very if riddled getServicesFromDelta?21:25
hatchugh I know! I spent forever stepping through various tests and it has pretty good code coverage21:25
rick_h_some of that is moved code I guess, are there real tests hitting that already I'm not looking at?21:25
hatchfrom a swath of different tests across the app21:25
rick_h_I know in my last branch I tried to go through and get things working, then did a -wip -cr so I could see the diff. Then tried to make sure I added/noted a test for each chunk in the diff (other than css/etc obviously) 21:26
rick_h_just a fyi, this is how I'm going through and reviewing is looking for the changes in one window, and looking for the tests in the others21:26
hatchright so the issue is that there is no way looking at the code to know where it's being testsed21:27
hatchwe need some way to figure that out21:27
rick_h_k21:27
rick_h_hatch: ok, replied. Got a LGTM with a note/two21:28
hatchthis relation code is tested from so many different integration tests21:29
hatchthis will be my new thing21:29
hatchto figure out a proper approach to this21:29
rick_h_ok, as I said. I'll take the word on it. I just get nervous as we had some recent relations breakages and such lately21:29
hatchrick_h_: the null is object test issue was with `typeof` not isObject21:30
hatchjust FYI21:30
rick_h_hatch: ah, ok. 21:30
hatchisObject has an isNull check21:30
hatchI looked ;)21:30
rick_h_hatch: well still, a check that you got an object seems a bit weak, but that's cool then. It was the big way I could see for it to pass wrongly. 21:31
hatchoh no that's all it needs to be21:31
rick_h_gotcha, k21:31
hatchyeah it's just a 'trigger' to tell databinding that it's changed21:31
hatchso I figured sending the change object made the most sense21:32
hatchas it will always be different21:32
hatchso 'change' will fire21:32
rick_h_right, but my first instinct was that you'd check that the object sent was the one received. 21:32
hatch(perfect place for a reviewer comment)21:32
* rick_h_ closed the window...sec21:32
rick_h_lol21:32
rick_h_ok, well I've got to go finish dinner. Hopefully that unblocks you. Didn't want to carry it overnight if I could help it21:33
hatchyup i'll add more tests with the new stuff tomorrow21:33
hatchbut smaller branches21:33
hatchsmaller branches21:33
hatchthe way of the future...21:33
hatchthe way of the future...21:33
hatchthe way of the future...21:33
hatchwell got the new vm up and running but man it's slow for some reason23:30
hatchappears to be related to the new UI effects23:33
bcsalleris the VM not GPU accelerated?23:38
hatchI'm not sure - it's doing the updates now, will see if that fixes it23:39
hatchI'm going to guess the 'parallels tools' didn't install properly or something so it's causing it to not be gpu accelerated23:39
hatchthose 'tools' on any of these vm providers never quite work properly :()23:40
hatch:)23:40
rick_h_in virtualbox you need to install the guest additions to get the gpu accel23:40
hatchyeah so that's my guess is that the tools are not working properly23:40
hatchother than that the OS feels way faster than 12.0423:41
hatchoh cool they now have a shell install script for Ubuntu23:44
hatchsomeone was thinkin23:44
hatchpfft ther we go 1Mx better23:46
hatchI really like the new juju.ubuntu.com23:57
hatchimho really gets the point across23:58

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