[00:48] jujugui lf a quick doc review https://github.com/juju/juju-gui/pull/37 [00:48] On it. [00:48] thanks [00:51] looks like I need to make some changes to get the docs to compile so I'll make any other changes you find too [00:53] hatch, +1 otherwise. [00:54] Makyo curious - it's under the Bit rotted instructions along with a lot of other stuff [00:54] should I just leave it? [00:55] Makyo: I'm out. I can qa again in the morning. I didn't see a reply back sorry [00:55] rick_h__, np, be well [00:56] Makyo: have a good night. The stuff in there ok? [00:57] rick_h__, Yep, just making sure it was basically LGTM otherwise, could submit tonight with those changes, on for another few hours. [00:57] Makyo: yea, sorry. I wasn't sure if there was back story to any of it and such. Like the escape things, the shortcuts used [00:57] I'll LGTM with comments [00:57] Cool, thanks :) [00:58] done [00:58] hatch, If the fix is quick, go ahead, otherwise, remove. [00:59] hatch: if those aren't system installed don't you have to find the path in the node_modules to run them? [00:59] e.g. lint and such? [01:00] hatch: I'm nervous about that change off the top of my head without matching updates to some bin scripts [01:00] rick_h__ when you install something with npm -g it's available to any node script [01:00] hmmm, you still -g, but it has to be sudo do write to the lib dir [01:00] not if you change the permissions [01:01] it installs in /usr/lib doesn't it? only root can write there [01:01] ugh, bad form to charm perms on /usr/lib man [01:01] usr/local/lib [01:02] and no, giving sudo to any node script on npm is bad form :P [01:02] hmmm, http://paste.ubuntu.com/6597268/ [01:02] I'm not a fan of this move [01:02] interesting... [01:03] rick_h__ you must be running an older install https://npmjs.org/doc/files/npm-folders.html [01:04] usr/local is the prefix now [01:06] hmm the python tests are failing [01:07] hatch: yea, this machine is still 13.04 I think. [01:07] hatch: but still, installing something globally in ubuntu is always a sudo thing [01:07] I mean version of npm/node when it was installed originally [01:07] I'm not a fan of changing that by messing with permissions in system level directories (outside of ~) [01:08] agree to disagree? ;P [01:08] heh, not without a card to talk about it on standup so I can get reinforcements :) [01:09] tbh I don't see where the disadvantage is here [01:09] and the test that failed was saucelabs [01:09] if anything, restricting it's permissions is better [01:09] it's monkeying with standard linux practices [01:10] the old issue you had to chown one directory in your home [01:10] this is playing outside of home, it's breaking multi-usre systems [01:10] user [01:10] if user rharding owns the global node modules, how can user mitechie install/use them [01:10] it's kind of fundamental to how this stuff works [01:10] if root owns them neither user can without sudo [01:11] right, but you can grant sudo to muliple users [01:11] and if sudo does something, everyone gets access and shares [01:11] you can also create a node-dev user group and assign it to them :) [01:11] this is setting things to one user, and the wrong one. It's a big breach of linuxy [01:12] fine, then update the instructions to use sudo to create a node-dev group, assign permissions to that folder to that group, and add your user account to that group :) [01:12] bath time, bbl, but veto without discussion on the change please [01:12] lol I think at that point the user knows what to do [01:12] ok I'll leave the PR open [01:16] and the user knew what to do the old way :) [01:16] plus tests still don't pass :P [01:19] hey thats your fault those are the sauce labs tests :P [01:19] ok I could do something like "Don't want random js scripts from running with sudo? Then you can do this...." :P [01:24] bah, it is a timeout isn't it. Crap [01:25] I had hoped my branch today would stop that. Oh well, going to production tomorrow should help [01:25] :) till tomorrow [01:25] cya [13:14] jujugui updating FF testing from 16 v16 to 26. Any objections? [13:16] actually 25, they don't have 26 === gary_poster|away is now known as gary_poster [13:24] * benji reboots to finish installing updates. [13:24] frankban, rick_h__ sorry, family issues made me start late. joining, frankban; may be late rick_h__ [13:24] gary_poster: np [13:25] gary_poster: rgr [13:55] hey y'all it's quiz time. who can guess why 'q' as a variable name is bad? [13:57] bac: because when you're in pdb you'll type "q" expecting to see its contents but instead you "q"uit the debugger [13:57] (not to mention that single-char variable names are almost always a bad idea) [13:58] benji: yes, you win a valuable prize! [13:58] benji: i've only done it three times this morning. (not a quick learner.) [13:59] heh [14:00] I've tried to train mayself to use "p q" and "! q = foo" all the time in pdb so as not to have that problem. I've not succeeded. [14:00] conflicts with other pdb commands are annoying but not quite as much as q. [14:27] I love the gmail mute-a-thread function. It's great when someone leaves the company or has a baby. [14:29] lol [14:34] rick_h__ did you see http://stridercd.com/ looks like a node version of jenkins [14:35] hatch: yea, and I talked with the dev about our landing issue [14:35] hatch: and he was interested in getting a plugin to do it [14:35] landing issue? [14:36] hatch: but it's not charmed, running nodejs (non-approved tech technically), and was a bigger learning curve [14:36] hatch: the jenkins-github-lander stuff. Automated github merge based on test run [14:36] ohh right right [14:37] so yea, I did look into it and talked with the dev. It's interesting, would have loved to had more time to follow that route as I'm not really a jenkins fan, but this was closer to working then strider for our needs [14:37] you probably didnt choose it because it was node.js lol [14:38] :P [14:38] Honestly, the fact that it would make it harder to get into general use and for juju-core played a factor [14:38] but personally, I went through the code and found it cool. [14:39] I wish we could have icons for unapproved charms [14:39] totally unrelated heh [14:40] but with these new must-have-tests rules my ghost charm will never get approced [14:40] approved* [14:42] jujugui I made some changes to the hacking doc which involve changing permissions on a /usr/local/lib dir instead of using npm with sudo, can you check it out and give your opinion plz https://github.com/juju/juju-gui/pull/37 [14:49] hatch: why is using sudo to modify directory ownership better than using sudo to install globally? [14:50] bac because when running npm under sudo your js script has root privileges - whereas the other way around it can only execute stuff your user can [14:51] hatch: excellent [14:53] We [14:53] We're not running our script under sudo ?.? [14:53] We shouldn't be, at least. [14:54] Makyo `sudo npm -g install jshint` executes the install script under sudo [14:54] so it can do what it wants [14:54] right, and we've selected jshint to be able to do things...like install a system (all user) available bin/jshint [14:54] Isn't that part of the contract of using npm? Same with apt-get, same with juju on a local provider... [14:54] it's the way it works. Just as you sudo apt-get install xxxxx does the same [14:54] or sudo pip install pyscss [14:55] it's the expected process. We're not doing "sudo npm install -g $user-supplied-thing" [14:55] If the node community doesn't have a framework for dealing with malicious packages in their management system, perhaps we should move away from node. [14:56] ok how about this.... I'll change it back to read with sudo then add "If you are not comfortable with giving npm install scripts root you can change the permissions of these directories..." [14:56] Makyo it doesn't anyone can upload anything to npm [14:56] Then perhaps we should move away from Node. [14:56] you can even override old versions on npm with new code [14:56] then add it to to anything installing python or system packages as well hatch [14:56] any time you install any software you're doing this [14:56] well I was under the assumption that the apt packages were reviewed [14:58] I was under the impression that npm packages were reviewed :) [14:58] haha, nope they are not [14:58] :T [14:58] I'm assuming the same with apt then? :) [14:59] Apt's reviewed. In order to make it in a release, a package has to be submitted and approved, thus juju-core requiring a PPA for a while. [15:01] well, main is. universe isnot so much. pypi is not reviewed [15:01] go get is definitely not reviewed [15:02] I've never run go get with sudo before [15:02] Hah [15:02] pypi requires sudo? [15:02] Outside of a virtualenv. [15:02] to install globally, which is why we use virtualenvs [15:02] what I would LOVE to see is using npm sans -g and building our scripts to use the node_modules dir [15:02] that's the *right* fix imo [15:03] this is just breaking linux for the sake of not doing ^ [15:03] +1 [15:03] I'm assuming we don't because it's a heavy install? [15:03] huh? no, we don't because -g means it's automatically on our system path [15:04] and otherwise you have to find the bin/xxxx in the node_modules path [15:04] at least that's my guess, I didn't write it out this way [15:04] gary_poster: call in a few? [15:04] bac yes, almost [15:06] gary_poster: ok, ping me here when ready. i'm there but not paying attention to it. [15:06] ack [15:20] bac, halloo [15:38] benji: got a sec? [15:39] rick_h__: sure [15:39] benji: I'm trying to see where the url the sauce labs tests hit gets set [15:39] benji: it looks like in browser.py it looks for $APP_URL [15:39] but I don't see it set anywhere [15:41] rick_h__: IIRC, the Jenkins config sets it when running "make ci-check" [15:42] bah, that's right. gotcha [15:50] jujugui call in 10, kanban now. [15:58] jujugui call in 1 [16:01] frankban, ping [16:08] Makyo: i'm interested to hear about your truck stop food. [16:08] bac, I convinced them to switch to Greek food, thankfully :D [16:08] a greek truck stop would be interesting [16:14] rick_h__, chrome and ff tests working, yay :-) [16:21] rick_h__ apparently Netfilix is going to start streaming 4k sometime soon....good for your new monitor ;) [16:21] assuming of course you don't have one of those life sucking isps which limit your bandwidth [16:39] hatch: :P [16:39] hatch: heh, are there any other kinds [16:40] and I don't watch tv on my monitors. I've actually got kind of low ens 720p tvs in the house [16:40] maybe one day I'll upgrade to a 'real' tv [16:40] hah, well it could be worse, we could life in the Orwellian future which is the UK's current internet [16:42] speaking of...curse you dell! still in 'production' [16:46] it'll probably be a while, it's very likely made in China, so it'll take a while to boat over here [16:51] ohhh d3 [16:52] Makyo we should probably update d3 we are a few versions back [16:52] hatch, agreed. [16:52] Will look into it; should be just a drop-in-place. [16:53] yeah we will need to make a new shrinkwrap so at the same time it probably wouldn't hurt to revisit the other deps [16:53] yui and the like [16:53] we HAVE to update YUI to 3.14.1 to support IE11 anyways [16:53] heh [17:01] hatch: heh, that answers a question about adding IE11 tests I guess then :) [17:01] rick_h__ hah yup [17:02] I mean to have us support it is pretty trivial upgrade [17:02] * hatch crosses fingers [17:02] heh [18:10] anyone here use znc? [18:10] hatch: I think jcsackett used to [18:10] I can't figure out how to connect to it heh [18:10] it 'should' work from what I can tell but I keep getting rejected [18:12] jujugui anyone care to chat and help me think through a decent way of bending the make/build system to my will of running make prod but with the debug config.js? [18:13] rick_h__, I can and would be happy to in a few--maybe 17 min. [18:13] gary_poster: rgr, appreciate it [18:13] np [18:13] if anyone else wants to do it that's cool too :-) [18:14] definitely, someone must have a tricky non fugly way to to this hidden in their head. All mine suck [18:15] rick_h__ make-prod --config debug then in the linked files have a switch :) [18:15] of course that would be a lot easier if we didn't use make [18:15] heh [18:16] * rick_h__ is going to come over there and cut your new powercord [18:16] hey it's a good idea! [18:31] hey rick_h__ do you have a hangout or hall I make one? [18:31] shall [18:31] gary_poster: can make one up, sec [18:31] cool [18:31] https://plus.google.com/hangouts/_/7acpjrujr1s6qjbhlf3kvuegro?hl=en [18:39] anyone know iptables well? I think my znc issue is that it's not allowing the data through the port https://gist.github.com/hatched/4c2effe2e92e9487bf13 can you see anything wrong with this output? [18:40] guihelp: I need two reviews + QA for https://codereview.appspot.com/44310043 (quickstart remove env). Anyone? no rush, thank you and have a nice evening! [18:41] frankban_: i'll do one later [18:41] frankban I will take one later [18:41] bac, gary_poster thanks [18:59] welp it's definitely a ip tables issue because I can't even get nc to connect [18:59] but lunch over, back to work [19:32] jujugui CI branch update review please. https://github.com/juju/juju-gui/pull/38 [19:32] two successful test runs in a row woot! [19:32] woot :-) [19:32] yay [19:32] on it [19:33] wow two in a row? That's gota be a first in MONTHS!!! [19:33] lol [19:33] hah! 10min run time atm with all three browsers only on build prod [19:33] added a card for adding back in build-debug, but was having timeout issues on that and prod is working nicely. Want to reevaluate once this proves stable-ish [19:34] rick_h__ looks like IE is last? [19:34] hatch: yes [19:34] We used to run it first because it was the most likely to fail [19:34] I'll add a comment but I wanted to hear if you had any input [19:34] well, I figure if it's last then you know the test runs itself was good, just IE hates you [19:34] if the first one blows up, you're not sure if it's CI in general or just IE failing a test [19:34] and the time diff of a coupe of minutes isn't going to make a big diff imo [19:35] rick_h__, +1 on prod only. we can move debug to aggregated post-commit perhaps, at some point (and maybe prod too if it becaomes too annoying) [19:35] I'll work on some instructions for running locally. I was running IE sauce tests from my local environment IE only so I think that's a "better" way to debug IE failures [19:35] hatch: so I'm -1 on moving it up the list imo [19:44] lgtmd [19:45] hah, poking at the github graphs. Go hatch go "On develop, 50 files have changed and there have been 343 additions and 1,302 deletions" [19:46] rick_h__ hah this is a little disturbing https://github.com/juju/juju-gui/graphs/contributors we remove more than 50% of the code we write! [19:47] bac has almost removed 100% of the loc that he has written lol [19:47] it's a good thing [19:47] hah, well that's some damn fine refactoring [19:47] lol is it? Aren't we supposed to do it right the first time....measure twice, cut once sort of thing ;) [19:47] though you're getting credit for removing the code I wrote. So wonder how that balances out over time. [19:47] hatch: nope, we're agile. We build, test, refactor, build, test, refactor [19:48] * bac rock! [19:48] haha [19:48] hatch, our last project would not allow you to add code unless you had removed an equivalent amount [19:48] haha oh really? Interesting [19:48] gary_poster: did you ever track that? [19:48] gary_poster: it happened as we were fading away [19:48] bac, true [19:48] orange squad had a bzr command we ran so we could justify crap [19:48] sucked because we kept claiming exceptions to get private projects finished [19:49] "I know this adds code, but private projects must go on....move aside" [19:51] anyone using clef, phone-based oauth tool? [19:52] nope, just lastpass mobile for that stuff [19:52] yeah lp me 2 [19:52] well, I guess that's a 2fa ish lastpass merged thingy [19:52] yeah, but clef automates the communication between site and phone by talking via squiggly lines [19:53] I don't get how it's 2fa though if it only requires your phone? [19:53] what's the second factor? [19:53] your FIST! [19:53] it's not 2fa afaict [19:53] Secure, easy, passwordless 2-factor authentication in less than 10 minutes. [19:53] from the website :/ [19:53] oh [19:54] seems like they are just an oauth service and they use your phone as identity rather than a password [19:55] rick_h__: i guess they consider the 4 digit passcode to get into the app as 1 factor [19:55] bac: ah, gotcha [19:57] it's interesting [19:57] I could see lastpass and that type of system working really well [19:57] although pattern is far more secure than 4 digit pin heh [19:57] rick_h__: i'm proposing the charmworld qa branch. would you have time to review it, being the godfather of all that stuff + migrations? [19:58] bac: sure thing, ok for it to be first thing in the morning? [19:58] rick_h__: yeah, i guess. charmworld isn't landing before monday [19:58] if then [19:58] I just got CI landed with IE support and want to drop the mic and EOD feeling all badass atm. :) and I have to run an errand to get presents in the mail atm [19:58] np [19:58] bac: thanks, I'll hit it up either later tonight or first thing in the morning. [19:59] jujugui all three browser saucelabs has landed. and if we want to add safari it's one step away as well. landing/test times are around 10min now [20:00] love it [20:00] * rick_h__ runs away before something breaks [20:00] thank you rick_h__ ! Very cool [20:00] very nice [20:02] cool [20:07] I'm actually surprised we didn't have CI for so long and everything we have still passes [20:07] how awesome are we??? [20:07] level of awesome: lucky [20:09] PFFT [20:14] very cool [20:20] benji i sometimes wish your irc comments came with a 'like' button. [20:21] That's how I'll make my millions: IRC + Like Button [20:21] hatch: remember, we're running what, 6 tests? [20:21] :P [20:21] i'd buy one.[*] [20:21] [*] requisite daily Raising Arizona reference [20:22] rick_h__: at your leisure https://codereview.appspot.com/44070045 [20:24] rick_h__ ohh you aren't running the unit tests in each browser? [21:03] I can't seem to write d3 code that looks good [21:03] no matter what chaining just looks horrible :/ [21:08] Makyo are you around? [21:08] Makyo when you get back from lunch I have a question about d3 not noticing a data change [21:09] What's up? [21:14] Makyo doh nm I screwed up, I forgot I was still in the d3 'enter' phase [21:14] sorry to botha [21:14] Glad I could help :D [21:15] haha rubber duck debugging :D [21:15] I'm switching the relations tab to use d3....wow there is a LOT more loc than just using a template haha