20:00:28 <david-lyle> #startmeeting Horizon 20:00:30 <TravT> o/ 20:00:31 <openstack> Meeting started Wed May 6 20:00:28 2015 UTC and is due to finish in 60 minutes. The chair is david-lyle. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:33 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:36 <openstack> The meeting name has been set to 'horizon' 20:00:50 <lhcheng> o/ 20:01:13 <absubram> o/ 20:01:30 <tqtran> o/ 20:02:01 <rhagarty_> hello 20:02:53 <david-lyle> letting people trickle in 20:03:06 <david-lyle> I think they're trickled 20:03:52 <david-lyle> I don't have any general announcements today 20:04:01 <david-lyle> I'm going to jump into the agenda 20:04:40 <david-lyle> #link https://wiki.openstack.org/wiki/Meetings/Horizon 20:04:55 <david-lyle> #topic Drop Django14 support and gate (lhcheng) 20:05:08 <lhcheng> hey 20:05:11 <david-lyle> hey 20:05:18 <david-lyle> got somethin' to say 20:05:20 <david-lyle> ? 20:05:37 <lhcheng> so rob found out that django14 support is going to be dropped oct 2015 20:05:45 <bpokorny> Hi 20:06:13 <lhcheng> with that, we should probably drop django14 support for L 20:06:39 <lhcheng> should we publicize it first in operators before doing it? 20:07:02 <david-lyle> we need to support the next LTS 1.8 before we drop 1.4 20:07:08 <david-lyle> but both should happen in L 20:07:57 <david-lyle> probably should publicize 20:08:16 <lhcheng> curious, why do we have to wait for LTS1.8? do we need to support last four releases? 20:08:51 <david-lyle> lhcheng: I think 1.5-1.6 aren't actually supported any more 20:09:01 <david-lyle> 1.7 is for now 20:09:27 <david-lyle> 1.8 is out, we just have to make sure it works with horizon 20:09:44 <david-lyle> (it doesn't) 20:09:49 <david-lyle> *spoiler alert 20:09:57 <lhcheng> david-lyle: lol 20:10:05 * david-lyle may have done that in the wrong order 20:10:10 <mrunge> in general, it works 20:10:11 <lhcheng> and the gate gate-horizon-python27-django14 moves to django17 then? 20:10:29 <david-lyle> lhcheng: yeah 20:10:48 <david-lyle> once we have 1.8 we could look at dropping 1.5 and 1.6 as well 20:10:58 <david-lyle> mrunge: just a few more issues to work out in d-o-a 20:11:09 <mrunge> david-lyle, yes- 20:11:20 <mrunge> we may rework our permissions system 20:11:30 <david-lyle> hmm 20:11:31 <mrunge> s/may/may need/ 20:11:50 <david-lyle> I meant to do that anyway 20:12:04 <david-lyle> with policy and has_perm in the same place now 20:12:18 <david-lyle> we can take that offline 20:12:26 <mrunge> yes 20:12:35 <david-lyle> any opinions on how many release we should stop supporting? 20:12:41 <mrunge> but this is something, I'm committed to 20:12:42 <TravT> mrunge: i'm interested in that as well. 20:12:53 <mrunge> cool! 20:12:58 <david-lyle> 1.7 introduces a lot of new features 20:13:02 <mrunge> sounds good 20:13:23 <david-lyle> spanning so many releases makes us do strange things 20:14:12 <mrunge> since we're gating on them, it's not that bad 20:14:19 <david-lyle> we'll start with fixing up 1.8, drop 1.4 and see where we end up on 1.5 and 1.6 20:14:45 <mrunge> sounds like a plan 20:15:07 <david-lyle> lhcheng: is there a bp to drop 1.4? 20:15:11 <lhcheng> david-lyle: sounds good, that question could also go with the django14 drop announcement 20:15:16 <lhcheng> david-lyle: don't think so 20:15:24 <david-lyle> sure 20:15:30 <david-lyle> let's add one 20:15:31 <doug-fish> does dropping 1.4 for Liberty affect anything for Juno/Kilo? 20:15:53 <mrunge> doug-fish, I wouldn't think so 20:15:56 <david-lyle> shouldn't, except certain backports will become trickier 20:16:10 <doug-fish> gate jobs? 20:16:12 <david-lyle> but since we're not adding features in backports 20:16:23 <david-lyle> the risk of conflict should be minor 20:16:28 <mrunge> david-lyle, it doesn't hurt us, if we're still gating on 1.4, right? 20:16:52 <doug-fish> I was wondering more if it mattered with the "drop 1.4" part of the plan 20:16:54 <david-lyle> mrunge: not in d-o-a IIRC 20:17:21 <mrunge> david-lyle, I mean, for juno 20:17:32 <david-lyle> shouldn't effect juno 20:17:37 <david-lyle> 1.4 job will still run 20:17:44 <lhcheng> mrunge: yeah, we should keep those job 20:18:09 <david-lyle> #topic Session data cleanup (lhcheng) 20:18:34 <mrunge> lhcheng++ on this! 20:18:44 <ducttape_> can we remove the "allowed" cache thing from the session? 20:18:47 <lhcheng> for the session data, we keep on hitting the cookie limit 20:18:57 <david-lyle> allowed? 20:19:05 <david-lyle> oh, for nav? 20:19:09 <mrunge> ugh! that hurts 20:19:09 <ducttape_> the cached permission thing, yep 20:19:12 <lhcheng> david-lyle: https://review.openstack.org/#/c/120790/ 20:19:30 <lhcheng> so I've been playing around to reduce the session size 20:19:53 <lhcheng> the "allowed" cache thing, takes 34% of the session size 20:20:11 <david-lyle> that's rediculous 20:20:11 <ducttape_> that's less betta' 20:20:13 <lhcheng> I suggest to move the caching from session to the django cache instead 20:20:19 <lhcheng> or maybe just drop it? 20:20:47 <mrunge> you know how often that's called per page view? 20:21:00 <david-lyle> should be once per page view 20:21:12 <david-lyle> but the number required should be dropping 20:21:14 <mrunge> without caching 10-20 times 20:21:31 <mrunge> I tried earlier today 20:21:51 <david-lyle> mrunge: you mean building the cache? 20:22:01 <david-lyle> or how many times the cache is read? 20:22:09 <mrunge> david-lyle, I mean, times read 20:22:11 <ducttape_> all the nav items in the layout stuff, are hiting that a bunch of times 20:22:34 <mrunge> yes, we should keep the cache 20:22:35 <david-lyle> we want to move to a model where we don't rebuild the nav on ever panel load 20:22:47 <david-lyle> only on rescope 20:22:57 <mrunge> sounds better 20:23:13 <ducttape_> what is the perf difference with that disabled mrunge? is it noticeable? 20:23:29 <mrunge> ducttape_, I can't really tell 20:23:33 <david-lyle> ducttape_: I really think it was for testing 20:23:39 <mrunge> I didn't measure that 20:23:45 <ducttape_> ok, thanks 20:23:52 <lhcheng> david-lyle: if the user rescope, this cache does not get reset. you just found a bug. :P 20:23:55 <r1chardj0n3s> in the interim, could a localstorage be used instead of django cache? just thinking of future nav work here ;) 20:24:29 <TravT> lhcheng: i just saw your comment and was starting reply on it in the patch. 20:24:35 <ducttape_> r1chardj0n3s: no, it needs to reside on server for current template to use 20:24:41 <TravT> on the angular side there is also caching 20:24:44 <TravT> once per page load 20:24:53 <r1chardj0n3s> ok 20:24:57 <david-lyle> r1chardj0n3s: I would believe that the nav component of the page will no longer need to reload 20:25:06 <david-lyle> thus the purpose of the cache would be gone 20:25:11 <r1chardj0n3s> david-lyle: that's the intention, yes 20:25:19 <TravT> but one of the things I tore my hair out over was how would I be notifiied when token rescope. 20:25:22 <ducttape_> lhcheng - you said you had another place to store that cache? 20:25:26 <tqtran> well, we can still use localStorage to store things like "last visited page" 20:25:28 <david-lyle> only the panel part of the page is loading 20:25:29 <ducttape_> someplace besides the session? 20:25:54 <lhcheng> ducttape_: yeah, just store it in django's built-in cache framework 20:26:13 <lhcheng> could be by process, file cache, memche 20:26:14 <ducttape_> and that would have a unique spot for each user / token then? 20:26:45 <lhcheng> yeah, just need to use a unique key for that user/token 20:27:30 <david-lyle> ok, I think we're all for reducing session size, we can discuss the particulars in a couple of weeks 20:27:32 <ducttape_> +1 lhcheng, our session stuff is a constant hassle. ship it 20:28:15 <mrunge> yes, session size issues are hard to debug 20:28:17 <lhcheng> so got the size drop by 75% on my local 20:28:25 <mrunge> cool 20:28:49 <lhcheng> david-lyle: sure, we can discuss the details later 20:28:55 <david-lyle> #topic Summit Scheduling (david-lyle) 20:29:03 <david-lyle> I wanted to make sure we have enough time for this 20:29:20 <david-lyle> session size/storage happens to be one of the proposed topics too 20:29:33 <david-lyle> #link https://etherpad.openstack.org/p/horizon-liberty-summit 20:29:39 <ducttape_> can we also talk about session size then? 20:29:46 <david-lyle> ducttape_: no 20:29:49 <david-lyle> :P 20:30:19 <david-lyle> let's talk session allocation 20:31:04 <david-lyle> someone did a nice job sorting through these, I assume either doug-fish or robcresswell 20:31:06 <david-lyle> not sure 20:31:18 <doug-fish> I took a shot at it 20:31:22 <david-lyle> thanks doug-fish 20:31:26 <doug-fish> sure np 20:31:45 <david-lyle> So doug-fish split out topics starting about line 37 20:32:04 <TravT> the topics you put at top are down to a single word, so now sure what is meant by some of them 20:32:18 <david-lyle> I need to publish a schedule so I started trying to figure out what are fishbowl and what are working sessions 20:32:35 <david-lyle> TravT: they reference the topics starting on 37 20:32:59 <david-lyle> my lazin^H^H^H^H^Hshorthand 20:33:26 <david-lyle> line 63, I don't think websockets applies 20:33:34 <david-lyle> to that topic 20:33:47 <david-lyle> I think it would go more with messaging 20:33:48 <mrunge> yeah 20:33:53 <david-lyle> if anything 20:34:09 <mrunge> it's just a q, if we want to go that route 20:34:20 <mrunge> there were so many issues to be solved with that 20:34:26 <TravT> there is a small aspect of websockets aspect to what I'd like to show / talk about with elastic search. 20:34:47 <TravT> but is not the main point, so separating them is fine with me. 20:35:40 <david-lyle> do people agree with the fishbowl topics? 20:35:44 <mrunge> I wouldn't think, *any* messaging infrastructure should take a whole session 20:36:09 <absubram> Hi all, I put down my request all the way at the bottom L-194 a couple days back.. not sure if anyone got a chance to look at it.. this is to request a better solution to some of the vendor specific code we have in the dashboard section.. 20:36:12 <TravT> so plugins is really broad. 20:36:14 <david-lyle> mrunge: I think better error messages and websockets could share a working session ? 20:36:22 <mrunge> do we want UX in a bigger plenum? 20:36:27 <mrunge> david-lyle, yes! 20:36:42 <absubram> Rob C will probably end up doing the work.. but we don’t have a solution just yet.. and we would like to talk about it.. don’t think it warrants an entire session to itself though.. 20:37:10 <TravT> david-lyle it seems like there could be a plugins from the context of which projects are core or not and that would probably take a whole fishbowl 20:37:11 <absubram> some time in the contributor meetup is fine too.. unless others have a different opinion? mrunge? :) 20:37:27 <david-lyle> TravT: absubram so for plugins 20:37:44 <david-lyle> I think it's a fishbowl and a worksession 20:37:51 <TravT> but does that cover tech discussion? 20:37:54 <TravT> ah, ok 20:38:13 <mrunge> absubram, that would be fishbowl number one 20:38:37 <absubram> ok.. that’s fine! 20:38:52 <david-lyle> I want to describe the proposed split and structure 20:38:52 <mrunge> david-lyle, do we have the possibility to have working sessions after fish bowls? 20:39:10 <david-lyle> in the fishbowl and then use a working session to plan on implementation 20:39:27 <david-lyle> https://docs.google.com/spreadsheets/d/1VsFdRYGbX5eCde81XDV7TrPBfEC7cgtOFikruYmqbPY/edit#gid=569963128 20:39:29 <TravT> david-lyle, which one does dependencies refer to? 20:39:41 <TravT> is that packaging? 20:40:22 <david-lyle> mrunge: looks like yes 20:40:27 <david-lyle> TravT: yes 20:41:09 <david-lyle> we'll have to see how the cross-project session goes, I suppose 20:41:58 <TravT> So angular 101 is really broad. 20:42:10 <david-lyle> TravT: yes indeed 20:42:21 <david-lyle> we need more than 1 angular session 20:42:50 <david-lyle> that one was the 101 to get people up to speed that haven't been neck deep in it 20:43:15 <TravT> yeah, it'll be tough to cover in 40 minutes that topic. 20:43:55 <TravT> Maybe the other topic could be an overview of our ideas for moving forward with angular in liberty 20:44:05 <r1chardj0n3s> good idea 20:44:17 <david-lyle> TravT: it's not a tutorial or class, just a pointer to where we are and how to get started 20:44:34 <david-lyle> OR we could move the angular chatter to Friday 20:44:41 <david-lyle> we have all day Friday 20:44:45 <r1chardj0n3s> david-lyle: should cover the styles, rules and layout we've developed for horizon 20:45:40 <david-lyle> maybe Friday morning is a better angular time? 20:45:57 <r1chardj0n3s> yeah, when we're all fresh ;) 20:46:04 <david-lyle> haha 20:46:45 <david-lyle> do we want a prioritization work session? 20:47:08 <mrunge> makes sense 20:47:10 <TravT> That was a good discussion in the contributors meetup in Paris 20:47:42 <TravT> As FYI, a number of HP'ers working on angular have to depart shortly after noon on Friday. :( 20:48:05 <david-lyle> friday afternoon then 20:48:07 <david-lyle> :P 20:48:11 <r1chardj0n3s> TravT: is that leave-the-venue depart or fly-out depart? 20:48:13 <TravT> :) 20:48:24 <TravT> leave the venue 20:48:33 <TravT> I have to leave the venue at about 2. 20:48:42 <r1chardj0n3s> ok, so we get a solid morning to hassle^H^H^H^H^H^Hwork with you 20:48:54 <david-lyle> Keystone could be meetup topic too 20:49:32 <david-lyle> giving merlin a session would be good 20:51:13 <TravT> mrunge put a topic up that would be good at some point. Moving forward with functions / deprecating old 20:51:41 <mrunge> TravT, yes! 20:52:03 <mrunge> TravT, more interesting: how to fix older releases then 20:52:19 <mrunge> since it's expected to fix in master branch first 20:52:23 <TravT> straight patch i guess. no cherry picking 20:52:29 <mrunge> we already discussed that 20:52:57 <mrunge> TravT, still that bears the risk of patching an issue in master-1 which is still existent in master 20:54:14 <mrunge> I wonder, if we really need a whole session for this 20:54:44 <david-lyle> I've been editing away 20:54:58 <david-lyle> numbering is not priority btw 20:55:16 <david-lyle> feel free to comment inline or pull something up 20:55:42 <tqtran> do we need a session for discussing karma and jasmine tests? 20:55:49 <tqtran> or just jasmine 101? 20:56:09 <david-lyle> i think that may be part of angular 101 20:56:39 <r1chardj0n3s> so that 40 minutes is filling up quite nicely then ;) 20:56:45 <TravT> tqtran: matt-borland also can help give a great overview of that and demo stuff related to karma 20:57:15 <david-lyle> r1chardj0n3s: moved it to Friday morning 20:57:19 <david-lyle> for continuity 20:57:25 <david-lyle> sake 20:57:48 <david-lyle> plus I have a feeling more people will show up than just the 28 or so the working session supports 20:58:07 <TravT> 28 people... that's small 20:58:13 <r1chardj0n3s> mmm 20:58:21 <mrunge> we should have a backup plan 20:58:22 <TravT> i guess we all better get in line and camp out overnight 20:58:22 <david-lyle> hence working session 20:58:29 <mrunge> a second topic 20:58:50 <mrunge> and just capture a different room 20:59:09 <david-lyle> times up 20:59:25 <david-lyle> I have yet to decide how descriptive the topics for the working sessions will be 20:59:33 <david-lyle> don't want too many people 20:59:40 <david-lyle> observing 20:59:49 <TravT> i think we had about 40 in the contributor's meetup 21:00:08 <david-lyle> I'll let the etherpad sit for another day or so then publish a schedule 21:00:24 <david-lyle> continue to comment on the proposed allocation of spots. 21:00:29 <david-lyle> Thanks everyone! 21:00:32 <david-lyle> #endmeeting