20:00:04 #startmeeting horizon 20:00:05 Meeting started Wed Apr 13 20:00:04 2016 UTC and is due to finish in 60 minutes. The chair is robcresswell. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:07 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:10 The meeting name has been set to 'horizon' 20:00:21 o/ 20:00:23 (◠‿◠✿)ノ 20:00:36 \o 20:00:44 o/ 20:00:49 are you left-handed bpokorny? 20:00:54 o/ 20:01:11 I forgot which, let me check :) 20:01:17 \o/ 20:01:23 lol 20:01:25 Ambidextrious 20:01:43 O/ 20:02:04 ¯\_(ツ)_/¯ 20:02:08 Couple of notices to begin with 20:02:16 #topic Notices 20:02:40 Gate stability should be back to normal levels, barring a couple of mischeivous integration failures 20:02:51 yey!!! thanks :) :) 20:03:08 We had a string of issues last week; literally 4 independent gate breaking issues in a row. 20:03:35 Row of misfortune 20:03:48 If there are any further issues, please ping me and let me know so I can look into it. 20:03:57 As if it was waiting for us to release :) 20:03:59 #info Gate stability back to normal levels 20:04:06 The second item is 20:04:14 #link https://www.openstack.org/summit/austin-2016/summit-schedule/global-search?t=Horizon%3A 20:04:24 This is likely going to be the final summit schedule 20:05:08 I've transferred most of the big discussion points and etherpad links to each one 20:05:45 If there's any mistakes or missing info, let me know. 20:06:09 Thats all from me, and the agenda is open. 20:06:15 robcresswell: btw, we tentatively fixed mischievous integration issues, https://review.openstack.org/#/c/305163/5 20:06:22 #topic Open Discussion 20:06:24 Infra is currently synchronizing new npm mirrors. They'll go into rotation early next week, which may add short-term instability in the gate, until the caches warm up. 20:06:39 A couple of more rechecks, and we'll be more confident 20:06:54 Thanks tsufiev 20:07:07 krotscheck: Thanks for the heads up 20:07:19 * david-lyle ended up on the wrong side of a net split 20:07:41 I think I info'd and link'd appropriately david-lyle :) 20:08:01 just reading the log :) 20:08:21 Upside: Leftpad won't take us down anymore :) 20:08:32 I'd like to remind about renaming AVAILABLE_REGIONS renaming couple of patches 20:08:47 tsufiev: link? 20:08:51 If we're to fix it, the sooner in cycle - the better 20:08:56 https://review.openstack.org/#/c/224756/ 20:09:01 david-lyle: ^^^ 20:09:33 wow 20:10:00 you're renaming the form fields too? 20:10:07 Yeah, some discussion may be... unavoidable :) 20:10:18 :) 20:10:26 Yes - to prevent further misunderstandings 20:10:26 will use the appropriate medim 20:10:29 *medium 20:11:15 Sure, gerrit will fit it better 20:12:03 krotscheck: Re: ESLint. What are your thoughts when two projects using eslint-config-openstack have opposing rules? Does eslint dictate the winner? Do we carry the warning and ignore it? Or does eslint do neither? 20:13:02 It seemed like Ironic UI and Horizon may have different styles regarding variable definition, for example (was trying to follow along with IRC) 20:15:21 robcresswell: I'm not certain in which context that would come up? When would eslint be run against two projects' code bases simultaneously? 20:16:17 krotscheck: Not at runtime, I meant with regards to which rules are carried by eslint-config-openstack 20:16:57 robcresswell: The extending configuration can override anything in the extended config. 20:17:27 So we'd just disable any rules we didnt agree with, and update the rest 20:17:49 robcresswell: The hope is that enough of the horizon team is on board of any rules that land, and that you remain engaged in maintaining the rule set. 20:18:02 robcresswell: See-> Hacking 20:18:08 Also, see-> pep8 20:18:29 Understood 20:18:33 Having big fights about code style is a waste of time. It's better to be consistent globally, so we can argue about things that matter. 20:19:32 That's true, but also easier to say when a one line rule change for you means a 500 line patch for us. Thats an exaggeration, but thats where I'm coming from. I'll make sure we stay involved in the rule reviews. 20:19:50 Easier than upgrading and pushing back later one. 20:20:37 global consistency can be overapplied 20:21:33 I can see the motivation though, openstack is bit of a pain to move across due to inconsistency. JS in one project can be a totally different style to JS in another, which just slows cross project work. 20:23:02 are there other JS projects in openstack beyond Ironic UI and Horizon? 20:23:26 I don't know if you count the plugins as separate projects 20:23:56 Because in that sense, yes, there are at least 3 or 4 Horizon JS plugin repos 20:24:12 there's also the app catalog? 20:27:19 quiet room 20:27:39 Yep, if nobody has anything they want to discuss we can call it early. 20:27:45 wait 20:27:45 I did 20:27:49 go ahead 20:27:58 Right now, in our markup 20:28:07 we don't have a general naming convention for classes 20:28:15 some of them are underscored, dashed or camelcased 20:28:33 +1 for dashes 20:28:43 makes it really difficult to go in and brand things, cause you have to go back and forth a lot to see which naming convention was used 20:28:47 I've been changing everything to dashes :) 20:28:54 matches bootstrap's convention 20:28:58 I recently put up a patch 20:29:13 and ducttape_ asked that we possibly deprecate the old class name 20:29:20 The issue mentioned the other day was that changing classes will break custom selectors in JQuery etc 20:29:25 #link https://review.openstack.org/#/c/300691/ 20:29:31 just wondering what the take on it was 20:29:37 I've just been changing them with my other patches 20:29:56 robcresswell: and custom js code in plugins 20:29:57 for anyone who's got a style built on the existing names, you're creating a great deal of work 20:30:00 i'd prefer something like PIE-chart_uSAGE 20:30:18 TravT ಠ_ಠ 20:30:22 LOL 20:30:29 * TravT ducks and hides 20:30:40 I'd like to clean it all up … but I"m open for any plan of attack for that 20:31:02 Most of the time, I completely rewrite the markup, so retaining the old class names is moot 20:31:19 oh, even better 20:31:48 My thought was to list the renames in release note. if you go through specifically renaming pie_chart_usage to pie-chart-usage, then a one line note is a lot better than half an hour of an engineer yelling at his CSS and punching his monitor. 20:32:01 in those cases, if you've created a theme, then your custom css should be inherited cleanly 20:32:05 hurgleburgler: if you write a letter to ML, you'll hear a lot of opinions ;) 20:32:20 Mostly from plugin devs 20:32:40 hurgleburgler: you'll need to write a migration script for everybody 20:32:49 it will need to handle all the cases 20:32:57 sed script, omg 20:33:00 ಠ▃ಠ 20:33:06 Generally if you're using theme variables you have less work, not more, with regards to CSS maintenance. 20:33:32 robcresswell: how many plugins do use them? 20:33:58 The pie charts? Or any custom js? 20:33:59 I guess time to start evangelizing it, meet the people 20:34:00 tsufiev: Most of them, because most plugins just use the Horizon templates. 20:34:12 Hmmm... 20:34:25 and most of the angular plugins are also up to date 20:34:43 angular uses their own pie charts 20:34:48 donut charts* 20:34:59 in that sense not much changes, if you've already got a theme in variables, then it will just clean itself up as hurgleburgler works 20:35:11 I think it becomes more difficult re: custom jquery though 20:35:19 hence the release note idea 20:35:37 Because if you're doing any form or chart manipulation, that would just stop functioning 20:35:42 Anyways, some preliminary warning should be made 20:36:30 I can add release notes and append to it as I go, when its needed 20:36:53 I would just like to point to the balance here 20:37:27 adapting for future/present ease vs creating a boat load of work for anyone who currently uses/depends on horizon 20:37:39 not that either are wrong 20:38:35 but lumping too many breaking changes in the present can effectively block a percentage of current users from moving forward 20:38:42 look at us with Django 20:38:54 and Django does a really good job of documenting those things 20:39:05 and deprecating things over several releases 20:39:22 * david-lyle stops pointing 20:39:36 They did a pretty good job of moving several things around undocumented too :( 20:39:37 I don't think these things are breaking changes though 20:40:13 hurgleburgler: you've written js and html extensions for Horizon, no? 20:40:16 Well, they are. If I write something that looks for pie-chart_usage and you rename the class, its broken. 20:40:33 I'm thinking beyond pie charts of course 20:40:42 Seems i missed something :/, got disconnected 20:40:50 Yes, I'm just giving a simple example. 20:41:02 If I make pie-chart-usage themed and use a different markup, then it breaks it anyways though 20:41:34 given, pie chart is actually constructing a pie chart on a given element using a class name, I think that's what makes it a special case though 20:43:18 Personally I think (in the scope of CSS renames) listing them is enough. We're talking about one grep|sed to fix it. 20:43:53 robcresswell: in an ideal world, but sure 20:43:58 k 20:44:36 When it comes to widgets, we should follow the existing process which has also been listing the component in the customising docs 20:49:22 Any other discussion points? 20:50:34 Thanks everyone 20:50:38 #endmeeting