20:02:26 <david-lyle_> #startmeeting horizondrivers 20:02:27 <openstack> Meeting started Wed Oct 14 20:02:26 2015 UTC and is due to finish in 60 minutes. The chair is david-lyle_. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:02:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:02:32 <openstack> The meeting name has been set to 'horizondrivers' 20:02:36 <robcresswell> o/ 20:02:41 <david-lyle_> #chair david-lyle 20:02:42 <openstack> Current chairs: david-lyle david-lyle_ 20:02:46 <david-lyle_> #chair robcresswell 20:02:47 <mrunge> o/ 20:02:48 <openstack> Current chairs: david-lyle david-lyle_ robcresswell 20:02:57 <robcresswell> In case of internet? 20:03:03 <david-lyle_> yeah 20:03:19 <david-lyle_> coming from two computers now, hopefully it won't be necessary 20:03:33 <robcresswell> haha 20:03:40 <mrunge> two handed chatting? 20:03:42 <TravT> o/ 20:03:49 <robcresswell> seems he spoke too soon 20:03:52 <david-lyle_> trying to find someone to agree with me 20:03:52 <TravT> david-lyle_: i feel your pain! 20:03:59 <david-lyle_> :) 20:04:12 <david-lyle_> General things, briefly 20:04:28 <david-lyle_> RC-2 is our final RC and will be the release for Liberty tomorrow 20:04:41 <TravT> |o? 20:04:55 <tqtran> [=_=]/3 20:04:59 <david-lyle_> Summit sessions are finalizing for Horizon 20:05:06 <TravT> i think i just accidentally invented a new emoticon 20:05:16 <TravT> (I have a question) 20:05:21 <david-lyle_> I like the hook 20:05:27 <david-lyle_> TravT, shoot 20:05:57 <TravT> oh sorry, no i don't have any question... was a pure accidental type 20:06:15 <david-lyle_> then yes 20:06:19 * TravT too many energy drinks, getting a little twitchy 20:06:38 <mrunge> you have a keyboard shortcut for: I have a question? 20:06:45 <robcresswell> haha 20:06:52 <robcresswell> The most uncertain person ever 20:07:00 * mrunge shuts up now. 20:07:35 <david-lyle> #link https://etherpad.openstack.org/p/horizon-mitaka-summit 20:07:39 <david-lyle> voice any concerns about the proposed schedule at the bottom by tomorrow 20:07:42 <david-lyle> as I will be publishing those 20:08:29 <david-lyle> those are the general items 20:08:36 <TravT> The angular fight club one will be tough for me to make at the 5:30 proposed time 20:08:38 <TravT> on Wed 20:09:11 <david-lyle> also fishbowl with ceilometer on Wed at 12:05 re: sorry state of ceilometer support in Horizon and what both teams can do about it 20:09:21 <mrunge> great 20:09:40 <david-lyle> TravT: other more pressing concerns at that time? 20:10:00 <TravT> Just that the main conference searchlight presentation I have ends at 5:20 20:10:19 <david-lyle> run ? 20:10:21 <david-lyle> jk 20:10:34 <TravT> and might be hard to rush over depending on if there are people who want to ask questions / talk after it 20:10:47 <TravT> i think it is in a separate building 20:10:51 <david-lyle> is hurgleburgler coming to Tokyo? 20:10:54 <TravT> yes 20:10:57 <TravT> :) 20:11:02 <david-lyle> ok cool 20:11:12 <david-lyle> theming wouldn't make much sense otherwise 20:11:29 <robcresswell> The one person in Horizon who understands out CSS right now. 20:11:32 <robcresswell> our* 20:11:33 <mrunge> will the fight club cover, why angular is so sloow? 20:11:37 <david-lyle> could swap theming/UX and 5:30 session 20:11:49 <david-lyle> mrunge: yes 20:11:53 <david-lyle> it better 20:11:54 <mrunge> awesome 20:12:16 <david-lyle> it's about adding more bottlenecks 20:12:38 <mrunge> do we have a list of bottlenecks to add? 20:12:56 <david-lyle> that's the slow down reason 20:12:59 <TravT> mrunge: that'd be great to bring information on "slow" with perf tests and analysis to a fight club 20:13:17 <mrunge> TravT, I added measurements to reviews 20:13:33 <TravT> oh cool. i hadn't seen them... 20:13:41 <TravT> but still bring them! 20:13:45 <david-lyle> we've gone from one set of API calls to services and added on top a large series of API requests to horizon over an even slower network 20:13:48 <mrunge> will do 20:14:20 <mrunge> even on a single machine with fast interconnection, it is slower 20:14:21 <david-lyle> anyway session fodder 20:14:25 <mrunge> yupp 20:14:38 <david-lyle> still over 2x the API requests 20:14:43 <david-lyle> ok, moving on 20:14:53 <david-lyle> or back to scheduling 20:15:52 <david-lyle> will move either theming or keystone into Wed 5:30 and shuffle the other working sessions down a slot 20:16:00 <TravT> thanks, david-lyle 20:16:06 <david-lyle> preferences on which? 20:16:30 <TravT> mine would be keystone... 20:16:38 <TravT> to go to wed 20:16:52 <TravT> but you, lhcheng, schedules for that would be key. 20:16:57 <david-lyle> doug-fish, lhcheng, ok with that? 20:17:12 <david-lyle> I have the master schedule here somewhere 20:18:00 <david-lyle> keystone PTL talk at that time, no keystone session 20:18:05 <david-lyle> lhcheng: is that an issue 20:18:06 <david-lyle> ? 20:18:35 <david-lyle> will track down later 20:18:42 <lhcheng> david-lyle: wed 5:30? 20:18:45 <david-lyle> yeah 20:19:17 <david-lyle> there seems to be a main summit keystone talk at that time, not keystone design session 20:19:39 <david-lyle> ok to talk keystone in horizon at that time? 20:20:16 <lhcheng> yeah, there is keystone fed and OSC design session at that time 20:20:29 <TravT> sounds like theming is better then 20:20:34 <david-lyle> ok theming it is 20:20:48 <lhcheng> yup, thanks for asking 20:20:54 <mrunge> yes, makes sense 20:21:04 <mrunge> but both topics are interesting... 20:22:12 <david-lyle> hurgleburgler: ok? 20:22:25 <david-lyle> going twice? 20:22:32 <hurgleburgler> will TravT not be able to make it? 20:22:48 <TravT> will probably at least be a bit late. 20:23:01 <hurgleburgler> k, then its alright with me 20:23:44 <r1chardj0n3s> o/ 20:23:52 <david-lyle> ok moved 20:23:53 <tsufiev> o/ 20:23:54 <TravT> r1chardj0n3s: o/ 20:23:56 <david-lyle> re-review 20:24:19 <david-lyle> now onto bps 20:24:26 <david-lyle> unless other concerns? 20:24:58 <robcresswell> I put a possible item on the agenda. Wondered if there was interest in discussing feature branching before the summit. 20:25:10 <TravT> robcresswell: agenda link? 20:25:28 <david-lyle> #link https://wiki.openstack.org/wiki/Meetings/HorizonDrivers#Agenda_for_October_14_2000_UTC 20:25:42 <david-lyle> a lot of that looks like summit topics 20:25:49 <robcresswell> heh, beat me to it. Thanks david-lyle 20:26:16 <robcresswell> The bps themselves are just angular bps as more of those folk around now 20:26:26 <david-lyle> but I'm open to starting the conversation now 20:26:42 <robcresswell> Cool! 20:27:10 <robcresswell> So the idea floated around was having a feature branch with more relaxed approval for a couple of the angular panels to get patterns and work together quicker. 20:27:27 <robcresswell> There are quite a number of people in HP/ Thoughtworks/ IBM being bottlenecked by our current process. 20:27:40 <robcresswell> And I'm worried by the idea of a repo split that has been mentioned as well. 20:27:59 <tqtran> I think we should also start the process for creating a feature branch, whether we end up using it or not. Streamline the process so that people arent waiting around an additional week or two until we can get their work onto the feature branch 20:28:09 <robcresswell> IMO, if community demands faster pace, we should meet that in a manageable way, which in my mind means feature branch. 20:28:45 <TravT> #link https://openstack.nimeyo.com/61606/openstack-horizon-suggestions-handling-refactors-future?qa_q=Suggestions+for+handling+new+panels&show=61606#q61606 20:28:53 <david-lyle> my concern is we're being inundated by changes that are very low priority if desirable at all 20:29:21 <david-lyle> while we're still trying to figure out a pattern 20:29:46 <david-lyle> feature branch is fine to try and arrive at that pattern 20:29:49 <robcresswell> david-lyle: Yes. The scope needs to be contained to 2/3 panels, not dozens. 20:30:07 <david-lyle> but things like system info holds no real value 20:30:23 <robcresswell> I would like us to have a feature branch for say, Users and Images to sort themselves out, then merge back later in M. 20:30:35 <david-lyle> it's a very seldom used page, duplicating in angular is not really necessary, especially at this point 20:30:49 <david-lyle> and provides no real advantage 20:31:12 <david-lyle> angular should be used to do the things it does well that django does not 20:31:31 <david-lyle> a static list of services is not dynamic, why treat it as such 20:32:04 * david-lyle try's to veer back from this tangent 20:32:34 <david-lyle> I'm fine with a feature branch for those items 20:32:38 <tqtran> the idea is, lets focus on 2-3 panels, have a feature branch for it, and have metrics that we can measure and a definitive date that we merge back into master 20:32:51 <robcresswell> Excellent. 20:32:53 <tqtran> whether it i perfect or not 20:33:30 <david-lyle> nothing's going to be perfect 20:33:43 <robcresswell> Right 20:33:44 <TravT> yeah, if we can't reach a milestone point for merging it back as "good enough", then we'll have to go to the new repo idea. 20:33:48 <tqtran> right, so for me, im all for the feature branch if those two conditions are met 20:33:49 <r1chardj0n3s> so the feature branch will have "more relaxed approval" does that mean we have to approve the merge back to master, ie. one ginormeous patch? 20:34:15 <TravT> r1chardj0n3s: i would think we would cherry pick some patches out 20:34:21 <david-lyle> by more relaxed, it means not a full replacement 20:34:21 <r1chardj0n3s> ok 20:34:26 <david-lyle> it's addititive 20:34:27 <TravT> maybe not "technicallly" cherry-pick 20:34:34 <TravT> more figuratively 20:35:00 <david-lyle> yes, new patches based on feature branch work 20:35:29 <david-lyle> but the idea is that those reviewing on the feature branch are comfortable with that code going into master 20:36:01 <TravT> a little clarification on that point... perhaps... 20:36:09 <david-lyle> for pure exploration, I suggest other options 20:36:15 <TravT> it can go onto feature branch to enable better collaboration 20:36:33 <TravT> but before going to master, it would be vetted with full tests, etc. 20:36:51 <david-lyle> those tests should be in the feature branch by the end 20:36:56 <TravT> agreed... 20:37:08 <robcresswell> I think you're defensively agreeing here guys :) 20:37:16 <TravT> david-lyle: other options? 20:37:23 <mrunge> should we squash patches then before merging back? 20:37:27 <tsufiev> TravT: and integration tests :) 20:37:32 <david-lyle> github? 20:37:43 <david-lyle> private repo? 20:38:16 <r1chardj0n3s> feature branches like http://docs.openstack.org/infra/manual/drivers.html ? 20:38:41 <david-lyle> r1chardj0n3s: yes 20:39:08 <david-lyle> although some projects have said it's easier to construct new patches from the content on the feature branch than doing a merge commit 20:39:18 <david-lyle> I think both methods have been used 20:39:18 <TravT> what do you mean by private repo? 20:39:19 <r1chardj0n3s> yep 20:39:39 <david-lyle> gerrit is not the only git 20:40:06 <david-lyle> why have all the check jobs run and fail on something you're not ready to propose? 20:40:42 <david-lyle> if you're looking for feedback, then fine 20:41:09 <david-lyle> I just think our signal to noise ratio is out of whack right now 20:41:25 <TravT> so private repo, do you mean a fork on github? 20:41:39 <tqtran> i think you guys are talking pass each other 20:41:43 <david-lyle> well that's more public 20:41:47 <mrunge> maybe gerrithub setup? 20:42:04 <mrunge> that would integrate a gerrit and a github repository 20:42:18 <mrunge> ... i.e a more upstream-like setup 20:42:25 <mrunge> instead of pull request 20:42:34 <david-lyle> I was thinking git init 20:42:36 <david-lyle> locally 20:43:07 <robcresswell> I don't think we need an external repo for the work we're suggesting. It's 2/3 panels in a feature branch, with slightly relaxed reviewing, so those being bottlenecked can collab better. 20:43:18 <david-lyle> robcresswell: not for that no 20:43:32 <robcresswell> Cool, the off-topic had me worried 20:43:50 <david-lyle> I just think we're getting a lot of half-thought through patches up 20:44:15 <david-lyle> it's ok to think them through further then propose, is all I'm suggesting 20:44:27 <david-lyle> if you're wanting to collaborate, then by all means 20:44:50 <robcresswell> awesome. So what's the next step for creating the branch? 20:44:53 * david-lyle hasn't checked the depth in pages of our reviews in a while and is scared to do so 20:45:07 <david-lyle> talking to relmgmt 20:45:19 <tqtran> we should start the process imo 20:45:29 <tqtran> doesnt hurt to have one ready when we decide to go with feature branch 20:45:36 <david-lyle> ok, will start the conversation 20:45:50 <david-lyle> #action david-lyle get that branch 20:45:50 <robcresswell> I agree, would be best to start that process sooner than later 20:45:59 <robcresswell> Woop, thanks david-lyle 20:46:27 <TravT> that will be a good process item out of the way. 20:46:51 <TravT> I'm quite certain david's point on the number of patchsets going through zuul was missed... 20:47:12 <TravT> i don't think feature branch changes that, but it should reduce rebases at least. 20:47:31 <david-lyle> sure, and if we're working collaboratively, it's ok 20:47:46 <david-lyle> zuul scales, most of the time 20:48:04 <tqtran> until end od release cycle lol 20:48:17 <david-lyle> exactly, most of the time 20:48:37 <robcresswell> Its mainly just that you often see people pushing test patches etc, using gerrit as a backup tool, which is just extra zuul load. People should be a little mindful of infra, thats all. 20:48:56 <TravT> so, next we have to agree on actual panels and a milestone targeting merge back... 20:49:08 <r1chardj0n3s> fight club! 20:49:09 <TravT> feature branch can't just be the new purgatory... 20:49:40 <tsufiev> robcresswell: people would be happy to be able to tell Zuul just relax 20:49:41 <tqtran> and a date as well.... 20:49:41 <david-lyle> no it's not, but a deadline is a guideline 20:50:46 <TravT> i wish we could have a wip on patches that stayed persistent and could also have one to say (don't test)... 20:50:46 <robcresswell> The panels is down to what angular devs are prepared to work on. Something like Users and Images, targeted for OpenStack M-3 (End of Fed/ Early March) 20:51:28 <TravT> m3 is too late. 20:51:31 <robcresswell> Release schedule if you're interested https://wiki.openstack.org/wiki/Mitaka_Release_Schedule 20:51:33 <david-lyle> I think common elements and API changes can come out when they are ready too 20:51:58 <tsufiev> btw, is anybody interested in writing integration tests for angular panels? 20:52:03 <robcresswell> TravT: It was just an arbitrary selection 20:52:19 <TravT> i'd like to consider searchlight in the mix... although i'm not sure that is better or not for me at this point. 20:52:19 <robcresswell> M-2 is mid January 20:52:42 <TravT> it probably is... 20:52:46 <robcresswell> TravT: That panel was awesome. If there is available devs, a third panel should be doable. But I would not go beyond that. 20:52:53 <robcresswell> Otherwise scope is unmanageable. 20:53:09 <TravT> the reason i'd like to consider it 20:53:28 <TravT> is i want to be able to inject searchlight search service into angular panels 20:53:36 <TravT> and for it to be able to inject actions from other panels into it 20:53:44 <tqtran> tsufiev: i would be, but im hung up on the users work atm. i want to get around to it though 20:54:32 <TravT> this probably won't make sense until i show people at the summit 20:55:03 <tqtran> gosh, after the summit people are going to go on vacations and holidays, it might be difficult 20:55:21 <tsufiev> tqtran: it'd be really good if someone volunteered, I'm a bit worried about integration coverage decline when we switch to angular panels from legacy ones 20:55:21 <TravT> i'm planning on sleeping 20:55:32 <robcresswell> TravT: No time for that 20:55:40 * david-lyle doesn't believe TravT 20:55:58 <tqtran> tsufiev: i agree fully, we should sit down and talk at the summit. im sure sqchen would be interested as well 20:56:01 * TravT david-lyle seems to know me 20:56:13 <tsufiev> tqtran: sure, will do 20:56:18 <TravT> matt-borland will be there... 20:56:30 <TravT> and the thoughtworks lead 20:56:30 <tqtran> and matt-borland, and anyone else hahaha 20:56:44 <TravT> maybe we can get some effort directed on that 20:56:49 <tqtran> just a brainstorm talk and then we will bring it up to the table and share 20:56:50 <r1chardj0n3s> tqtran: actually, I'm moving house the day after I get back from the summit; I clearly do vacation wrong ;) 20:57:05 <tqtran> r1chardj0n3s: omg lol yes definitely 20:57:20 <robcresswell> I'm taking the week off to wander round Japan :D 20:57:25 <tqtran> me too! 20:57:27 <robcresswell> post-summit 20:57:27 <TravT> so, should one of the working sessions be dedicated to testing? 20:57:43 <david-lyle> maybe part of the collab meetup 20:57:43 <TravT> maybe even r1chardj0n3s and tsufiev giving a talk on it? 20:57:58 <hurgleburgler> +1 20:57:59 <TravT> educational kinda thing? 20:57:59 <tsufiev> Hmmm 20:58:02 <david-lyle> we don't have topics work items penciled in 20:58:36 <r1chardj0n3s> I guess that could be done. What's the missing info in people's testing knowledge? 20:58:47 <tqtran> try everything lol jk 20:58:49 <david-lyle> moar tests 20:59:13 <tsufiev> TravT: I could prepare some integration tests how to 20:59:13 <TravT> r1chardj0n3s: how do you describe what you don't know? 20:59:14 <david-lyle> an integration testing strategy around angular content? 20:59:18 <tqtran> those mocks sean mentioned are starting to sound real nice :P 20:59:24 <TravT> i mean, i don't know what i don't know 20:59:35 <david-lyle> more than you know 20:59:43 <robcresswell> tqtran: Both of the mock bps are on the agenda for this meeting lol :p 20:59:49 <tqtran> not that you know 20:59:58 <TravT> what are we talking about again? 21:00:01 <tsufiev> david-lyle: I'm definitely not in the angular party :/ 21:00:12 * TravT trying to make jokes, not sure if they are coming across in irc 21:00:15 <david-lyle> party? 21:00:22 <tqtran> tsufiev: i have prepared an invitation for you already, its coming.... 21:00:31 <robcresswell> This is getting very surreal 21:00:51 <tsufiev> Haha, it' say bad English 21:00:59 <david-lyle> well I'm glad we had this time together 21:01:02 <david-lyle> :-D 21:01:04 <tsufiev> I meant 'a group of people' 21:01:11 <TravT> tsufiev: what are you saying? 21:01:16 <david-lyle> I thought political party 21:01:20 <tsufiev> Yep 21:01:26 <robcresswell> I have no idea what's going on 21:01:30 <david-lyle> time's up 21:01:35 <david-lyle> thanks everyone 21:01:38 <david-lyle> #nedmeeting 21:01:43 <robcresswell> loool 21:01:43 <david-lyle> #endmeeting