=== Makyo is now known as Makyo|out === rogpeppe1 is now known as rogpeppe [11:35] teknico, frankban: could one of you review my wee branch? [11:35] and, good morning. :) [11:36] bac, hi, do you mean your wee hours branch? :-) [11:36] teknico: no. my "wee" as in 'not many lines' and 'horribly boring' [11:37] bac, it's https://codereview.appspot.com/6873055 , right? [11:37] teknico: yep. sorry i wasn't specific, but it is linked off the card in the review lane [11:39] bac: on it [11:39] ok, but i only need one review. y'all can fight it out! [11:39] bac, it's settled already :-) [11:46] bac: approved. [11:46] frankban: thanks [11:57] good morning [12:02] hazmat: morning. hey where are you located these days? [12:11] bac, dc still. costa rica ( west coast beach town) move out on dec 13 [12:11] well.. technically va.. moved out of our dc house already [12:11] was up early (and late last night) doing support on the gui charm [12:12] hazmat: cool. look forward to seeing palm trees and toucans in the background during hangouts soon. [12:13] * hazmat pictures himself hunting jaguars and ocelots from a zipline with a bow and arrow [12:13] hmm.. hard to hold a martini that way [12:14] ah.. camelbak ;-) [12:29] bcsaller, ping [13:08] bac I have a conflict with our call on Thursday. Would moving our call an hour earlier be OK with you? [13:08] gary_poster: that's fine [13:08] thanks [13:09] one off or permanent? [13:09] one off bac [13:09] ok [13:25] mattuk1972: did you have any CSS changes you wanted to see for the charm store filter besides better placement of the checkmark? [13:37] bac, just removal of the 1px outline [13:43] mattuk1972: i'm looking but don't know exactly what you're referring to regarding the 1px outline [13:45] mattuk1972: is it the border: 1px solid #a3a1a1; [13:52] Makyo|out, you can't start bug 1083545 until teknico completes bug 1078910 [13:52] <_mup_> Bug #1083545: Juju GUI charm: serve the GUI using static files. < https://launchpad.net/bugs/1083545 > [13:52] <_mup_> Bug #1078910: app directory should work with simply serving files < https://launchpad.net/bugs/1078910 > [13:52] AFAIK [13:56] bac, yes [13:59] gary_poster, Makyo|out, which, with the valiant help of frankban, is almost ready for review [13:59] great teknico [14:05] gary_poster, teknico_ , frankban - I don't think it's totally blocked. I've been working on just getting nginx in there with the right location directives, though you're right that it won't serve properly until then. [14:05] Makyo|out, cool, sounds great [14:07] Makyo|out, yeah, that's great === teknico_ is now known as teknico [14:13] mattuk1972: better wrt outline and check placement? https://dl.dropbox.com/u/420990/Screen%20Shot%202012-12-05%20at%2010.12.51.png [14:14] gary_poster: ready when you are for our call, I am in the hangout, when you have time [14:18] bad, looks lots better ty === bac is now known as bradleyalexander [14:19] mattuk1972: better is good. [14:19] mattuk1972: more suggestions before i land it? === bradleyalexander is now known as bac [14:20] bac, just the text alignment then its all good [14:21] mattuk1972: what do you want to see changed, exactly? [14:22] bac, the text should be indented from the left edge - the same as the other drop down and how the same as the text appears in the drop down before click. [14:23] mattuk1972: ok [14:27] mattuk1972: looks like padding-left: 4px does it. what do you think? https://dl.dropbox.com/u/420990/Screen%20Shot%202012-12-05%20at%2010.27.10.png [14:35] bac, if that was 4 then 6 will make it the same as the drop down text b4 click. could you add the same padding to the bottom of it too? so the text isn't so close the the bottom edge? [14:38] mattuk1972: can we do a quick google hangout? [14:42] brb === Makyo|out is now known as Makyo [15:19] Makyo: i see your rev 264 doesn't show any reviewers. i wonder why that happened. [15:25] hazmat, benji and I cannot duplicate the problems that you encountered with his branch. We tried simple make and we also tried following the steps that uistage followed. the pastebins show you using sudo make, and we don't know why you would want to use sudo (http://paste.ubuntu.com/1411255/). Maybe sudo was a reflexive debugging thing from an earlier problem? [15:26] And do you have any other thoughts on duping? [15:26] we simply don't support jitsu deploy-to or make server yet in the charm, fwiw [15:27] gary_poster, try deploying to an env and add unit, wait for one to fail [15:27] hazmat, with that branch, not with trunk, right? [15:27] gary_poster, jitsu deploy-to doesnt have a delta for connecting to the env [15:28] or executing a charm [15:28] gary_poster, i was on trunk revno 265 [15:28] oh.. for the previous failure.. revno 263 on trunk i think [15:28] trying it with the branch should do it, if the problem lies there [15:29] bac bcsaller benji frankban goodspud hazmat Makyo mattuk1972 teknico call in 2 [15:29] mattuk1972: the changes you requested have been landed [15:29] 1 now [15:30] bac, nice one [15:30] mattuk1972: i noticed the expanded widget was shifting to the right. i changed that to make it drop straight down [15:30] hazmat, right. and then the dupe instructions are not "wait for it to fail" because that happens @ current trunk [15:30] cool [15:31] gary_poster, different failure though [15:31] ok [15:31] gary_poster, its hard to say definitively since the output is obscured [15:31] ack [15:31] gary_poster, but the previous one was a hard one... that manually running the make would also hit.. the current issue seems to more intermittent [15:31] ie manually running the make can succeed [15:32] right make works on dev box [15:48] hazmat, "recursive make considered harmful", really? had no idea [15:53] teknico, it was a bit tongue in cheek [15:53] Makyo, sent info on hr stuff via priv msg [15:53] hazmat, sphinx has its own makefile, can't see a better way :-) [15:54] ah.. [15:54] good point [15:54] hazmat, but feel free to find more problems with that branch ;-) [15:54] pls consider that a non review then [15:54] dig on the make help output though, that's great [15:55] hazmat, not all the available targets are described, only the most useful ones [15:56] tried to strike a balance between readability and completeness :-) [15:57] also, resisted the temptation to reorder and overhaul the whole Makefile :-) [16:04] teknico, if you have a plan on reordering/overhauling that you think you could do in < day, +1 [16:04] (as separate card) [16:05] you would ideally coordinate with Makyo and benji, who are in the same world, so that your changes would not make their lives inordinately difficult [16:06] I have resisted the same urge a couple times recently. :) [16:07] benji, are you trying to dupe what Kapil mentioned above? If we can get this resolved quickly that would be super [16:07] gary_poster: yep; no luck yet [16:07] do you want to pair for just a bit benji? [16:07] up to you [16:07] sure [16:07] ok [16:07] I managed to deploy the charm, but it is stuck in pending [16:10] bac bug 1086790 would be super duper. causing pain [16:10] <_mup_> Bug #1086790: charm does not include makefile stdout or stderr

< https://launchpad.net/bugs/1086790 > [16:11] if you are looking [16:24] sorry, wind+hale storm going strong here [16:31] teknico did you see my comments about makefile? [16:31] teknico_, ping? [16:32] gary_poster, back, hopefully for a while [16:32] frankban, teknico_ am I right that the charm builds the gui as root? [16:32] teknico_, did you see my comments about the makefile? [16:32] specifically: [16:32] teknico, if you have a plan on reordering/overhauling that you think you could do in < day, +1 [16:32] (as separate card) [16:32] you would ideally coordinate with Makyo and benji, who are in the same world, so that your changes would not make their lives inordinately difficult [16:33] gary_poster, yes, I saw those (just before being thrown out) [16:33] cool [16:34] gary_poster, better discussing it before moving forward, it seems :-) === teknico_ is now known as teknico [16:34] teknico, invited you to hangout, you see? [16:35] gary_poster, in juju-ui? getting there [16:35] teknico, no [16:35] getting url... [16:35] oh, ok [16:35] teknico, https://plus.google.com/hangouts/_/8ee850f7066069eb3f8c5de034cb29a9da690f7d?authuser=1&hl=en-US [16:41] gary_poster: re who build the gui in the charm: the user executing the hook, i.e. root (IIRC) [16:42] bcsaller or hazmat, am I right that charms are run as root? [16:42] frankban, exactly, thanks [16:42] they do [16:42] thanks bcsaller [16:47] Makyo, thanks for the review, are you totally positively sure there's nothing at all that can be improved (or just complained about :-) )? [16:55] gary_poster, they are [16:55] gary_poster, but its still good to put in the sudos [16:55] thanks [17:02] thanks for the review Makyo [17:22] teknico, frankban, Makefile is kind of a mess, but I agree with gary_poster in that it would probably do well as a separate card. [17:23] Sorry for delay, fighting with charm :/ [17:56] benji, gary_poster, previous makefile behavior restored as agreed, with hints added to various places [17:56] cool [17:56] cool teknico [18:00] Oh man, Dave Brubeck died today :( [18:00] * Makyo switches music. [18:37] hazmat, I'm sorry you had to stay up last night getting the charm to work. In the future, if you have a command performance of something we are working on that you have to deliver, please feel free to pass it to me/us to get it working. I understand these things come up suddenly sometimes, and we can try to deal with what we have, and we might know more about the status of the parts in motion. And even if not, we [18:37] can help. [18:37] benji, how is the charm doing? [18:39] gary_poster: "agent-state: started" [18:40] gary_poster, no worries .. it was out of the blue [18:40] as in they contacted me yesterday at 5pm est for help [18:40] * benji returns to lunch. [18:40] heh, lovely hazmat. [18:40] thanks benji [18:41] gary_poster, it happens.. i felt sorry for them.. they where in the london office till past midnight i guess [18:41] gary_poster, where you guy able to reproduce? [18:41] * hazmat needs a permanent auto replace for where were [18:42] hazmat, heh. summary is that we have not duped, and we think we have a story for how everything resolves to the intermittent failure bug that you filed later. Details: [18:43] - we used trunk as of benji's branch for all tests. For the charm, we pushed it to LP separately since we don't have a revno control in the charm [18:44] - running make locally worked always, including after partial cleans, such as of the way build works. [18:44] - to try and dupe failure in uistage, I incrementally moved through revisions to benji's [18:44] the first reload of benji's gave the error I saw at uistage [18:44] but then a subsequent reload was ok [18:45] hmm [18:45] it turned out it was because of a cached index.html asking for old locations of css [18:45] and then the new index.html was correct [18:45] - for the charm, we did it twice [18:46] the first time it failed, and we were bitten by the "charm eats stdout" problem as to diagnosis [18:46] but it failed in make [18:46] and when we ran make again, it worked [18:46] then we did make clean && make and that worked too [18:46] For the second run [18:47] we hacked the charm to no longer eat the stdout [18:47] and then started it back up again [18:47] this time it worked fine, per benji's report above [18:48] the pastebins that you gave at the time, at least all the ones that we looked at together, were from various surprises [18:48] such as the fact that make is run as root [18:48] but then you tried to run make as ubuntu [18:48] which failed [18:48] right.. ignoring that one [18:48] and then subsequent makes are hosed [18:49] e.g. http://paste.ubuntu.com/1411255/ [18:50] So in conclusion, and to repeat my summary, I think we have explanations for the symptoms we saw. I think we should re-land benji's branch, and then land a change to no longer eat stdout/stderr [18:50] and keep on with other bugfixes [18:50] gary_poster, sounds good [18:50] gary_poster, benji thanks for digging into it [18:50] cool, thanks hazmat [19:04] gary_poster: I'm relanding my branch now [19:04] great, thanks benji [19:04] benji, uistaging ;-) [19:05] gary_poster: if that is a reminder to use the QA lane, gotcha [19:06] it is, benji. That was the "if you make it just hard enough to figure out then maybe people will think about it and remember it better" approach. I will leave it to your imagination as to whether or not that was intentional, or merely a nice excuse for misremembering "uistage" :-P [19:07] heh [19:09] just added 8 units of the gui in ec2.. all came up nicely [19:34] good [19:37] * benji reboots in the hope of getting Chromium to work again. [19:43] my machine was so much more stable on quantal [19:52] lp2kb does not like the 'prototype' lane. if it is 'in progress' then you must be coding, dammit [19:56] Running on nginx :) http://ec2-23-22-181-236.compute-1.amazonaws.com:8888/ [20:05] we need a /version page (or similar) so we can tell which version is running on the stage [20:08] ok, I am declaring the un-revert to be done; the uistage is still working 15 minutes later [20:24] benji, ping.. [20:24] hazmat: yo [20:24] benji, i'm trying to use tb.. i get a Can not authenticate with server... any ideas? [20:25] hmm, that's not one I think I have seen; are you using two-factor auth? [20:26] benji, nope [20:26] this is when running "tb send ..."? [20:27] benji, recieve.. nm.. its working now [20:27] benji, it seems it doesn't work with the auth file [20:27] benji, but cli passing works okay [20:27] hmm [20:27] maybe I broke the config file handling recently, that or you changed your password in your sleep and don't remember ;) [20:33] benji, not sure.. created a new auth file.. verified password. dunno [20:33] passing the cli flags for user/pass worked so.. that's good enough [21:24] bac: is it still the case that LP staging doesn't do two-factor auth right? [21:24] it does benji. I had a comment about that in the doc I wrote in that branch you are finishing [21:25] cool, looking [21:25] AIUI, either you get a separate two-factor auth for staging (which I don't actually know how to do) or you ask the webops to turn off the two-factor auth for you in LP staging. [21:26] I recommend the second because It Worked For Me [21:38] gary_poster: what did we decide about development releases? Should they be named "VERSION-NUMBER-LAST-RELEASED-dev-r123" or "unreleased-dev-r123" or something else like "development-release-r123" [21:40] benji, I don't remember the decision. On the face of it, I prefer something like the last two (even "dev-r123"). However, I was concerned about the fact that the connection between the dev and stable releases would not be clear. you might have had a reply/reassurance about that? [21:42] IOW, the fact that dev-r123 comes after 0.5.0 and before 0.6.0 is not clear, and I like the usual naming schemes because they are clear in that way. I don't have a concrete use case for it though atm [21:42] gary_poster: my (small) concern is that using the last release's version number can be somewhat misleading; for example if we are working on what will become 1.5.0 but the last release was 1.4.6, then a non-final release would be 1.4.6-dev-r123; that makes it look like it is just a bit different than 1.4.6 instead of "almost 1.5.0" [21:43] benji, before my call at 5 I should talk with you about teknico's branch. You and I are both concerned about similar things, but I have a bit of a different take. [21:43] gary_poster: now is good for me [21:43] benji, yes, agree, but that's because I am acquiescing to your dislike of guessing what the next release number is :-P [21:43] :) [21:44] benji, juju-ui is open [21:44] If you are cool with the 1.2.3-dev-r123 versions, then I'll go with that [21:44] coming over [21:45] benji: i'm unsure about lp staging and 2fa [21:45] bac: you have to have it disabled, that worked well for me [21:45] since they made it mandatory i figured they had to have some plan... [21:45] for some definition of "mandatory"