20:00:23 <robcresswell> #startmeeting horizondrivers
20:00:24 <openstack> Meeting started Wed Jun  1 20:00:23 2016 UTC and is due to finish in 60 minutes.  The chair is robcresswell. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:27 <openstack> The meeting name has been set to 'horizondrivers'
20:00:32 <tyr__> o/
20:01:01 <robcresswell> Ha, quiet room :p
20:01:18 <tyr__> didn't this happen last week? :)
20:01:19 <r1chardj0n3s> o/
20:01:20 <lcastell> o/
20:01:33 <tsufiev> o/
20:01:42 <bpokorny> o/
20:01:45 <ediardo> o/
20:01:49 <robcresswell> Yeah, meetings seem a little quieter these days
20:02:33 <robcresswell> We're pretty light on agenda today, so will just give a couple notices then we can have an open discussion
20:02:59 <robcresswell> The first milestone for Newton will occur today/tomorrow
20:03:22 * robcresswell finds link
20:03:31 <robcresswell> https://launchpad.net/horizon/+milestone/newton-1
20:04:06 <robcresswell> I'll bump the remaining items to N-2 and we'll progress from there. Blueprints are a little behind where we'd like, but most are either in flight or have had several patch sets now
20:04:28 <robcresswell> Huge number of bugfixes though, so thats great :)
20:05:12 <robcresswell> The stable maint release for mitaka will also go out soon, waiting on a couple of merges. Good number of backports. Remember to tag anything you'd like backported with 'mitaka-backport-potential' or ping me about it.
20:05:26 <robcresswell> Finally, a big thankyou to tsufiev for working on the gate issues
20:05:40 <robcresswell> #topic Open Discussion
20:06:01 * tsufiev hopes it won't happen again
20:06:22 <r1chardj0n3s> I think we all hope that, but we all know the gate too well :/
20:06:35 <r1chardj0n3s> we still don't have upper-constraint limiting in
20:06:36 <tsufiev> the hero of this week is npm-run-test :)
20:06:44 <r1chardj0n3s> yeah
20:07:08 <tyr__> brief plug for https://blueprints.launchpad.net/horizon/+spec/angular-template-overrides and the associated patch https://review.openstack.org/#/c/323534/. I'm looking at pre-loading the Angular template cache which also allows the theme to override templates if needed.
20:07:59 <robcresswell> tyr__: Woah; does this solve the template overrides too?
20:08:01 <r1chardj0n3s> upper constraints patch reminder https://review.openstack.org/#/c/290203/
20:08:14 <tyr__> hurgleburgler convinced us that without Angular template overrides, people would feel like client-side views weren't as extensible as Django ones.
20:08:38 <r1chardj0n3s> seems reasonable
20:08:41 <robcresswell> I would agree with that. That looks like a really concide solution! 90 lines.
20:08:53 <robcresswell> concise*
20:09:05 <robcresswell> Thanks for working on that tyr__
20:09:36 <tyr__> working on cleaning it up a bit, but the basic idea appears to work. If you have opinions, I'd love to see any comments on the patch.
20:09:48 <robcresswell> Starred both of those. Bit under the weather atm but will get to it tomorrow
20:09:59 <david-lyle> question re: upper-constraints
20:10:24 <r1chardj0n3s> david-lyle: yup?
20:10:32 <david-lyle> I am still having problems with the tempest plugin. Now I think it's not even part of the build image
20:11:04 <david-lyle> It will work soonish, but do we want to pull the integrated tempest tests now? or still wait for the plugin to work?
20:11:26 <david-lyle> this will unblock the xstatic packages
20:11:28 <r1chardj0n3s> we don't *need* the tempest change for us to use upper-constraints, but it will prevent us from breaking tempest if we ever release an xstatic ;-)
20:11:40 <r1chardj0n3s> but yeah, just pulling the tempest test would also work
20:11:54 <r1chardj0n3s> since it's only one test
20:12:05 <r1chardj0n3s> but it would be good to get all the other xstatic publication stuff moving again
20:12:21 <david-lyle> I don't think there is a big risk
20:12:42 <david-lyle> I have the tempest patch up, I just need to de -1 the workflow
20:12:54 <tsufiev> do we have these tempest tests documented somewhere - the scope and purpose etc?
20:12:56 <robcresswell> r1chardj0n3s: Are we still waiting ona spec?I thought I recalled seeing one floating around in infra
20:13:14 <r1chardj0n3s> robcresswell: spec for the tempest change?
20:13:18 <david-lyle> tsufiev: it's one test, that verifies we can log in
20:13:26 * tsufiev is having problems catching up all the new stuff in horizon
20:13:29 <david-lyle> there, documented
20:13:36 <robcresswell> r1chardj0n3s: xstatic changes
20:13:36 <david-lyle> :P
20:13:39 <tsufiev> ah, okay
20:13:40 <robcresswell> releasing etc.
20:13:59 <david-lyle> the test moved to https://gitub.com/openstack/tempest-horizon
20:14:06 <r1chardj0n3s> robcresswell: hurm, there's a bit of docco around, I'll have to find it all again ;-)
20:14:18 <r1chardj0n3s> some specs were written, but they were outdated by the tempest change
20:14:23 <r1chardj0n3s> because everything got so much easier
20:14:34 <r1chardj0n3s> once we didn't have to worry about breaking that
20:15:13 <david-lyle> I'm going to reset the workflow on the tempest patch, and then work to get tempest-horizon into the devstack disk image
20:15:22 <david-lyle> unless someone protests
20:15:43 <david-lyle> we want xstatic updates as soon as possible
20:15:43 <robcresswell> r1chardj0n3s: Understood. I need to sit down and check through it all again
20:15:52 <r1chardj0n3s> robcresswell: I need to also :-)
20:16:12 <robcresswell> Thanks david-lyle
20:16:27 <robcresswell> I have a few more dependencies to release once the risk of breaking goes away
20:16:27 <ediardo> guys: I noticed that "More complete server side filtering" is marked as a High Priority for newton, can you check a BP that tries to address that part? https://blueprints.launchpad.net/horizon/+spec/server-side-filtering
20:16:55 <robcresswell> I'm reluctant to do anything that drives much content (like angular) until we are just breaking ourselves and not everyone
20:17:22 <robcresswell> ediardo: Reading now :)
20:17:51 <ediardo> there are ~28 tables using some sort of filtering, only 3 are using server-side
20:18:04 <robcresswell> Thats a very big blueprint scope
20:18:41 <david-lyle> robcresswell: would you prefer per table or service?
20:18:54 <robcresswell> I'm curious, how are you planning to unify the experience?
20:19:18 <david-lyle> unify?
20:19:21 <tsufiev> by establishing a new cross-project API group :)?
20:19:38 <robcresswell> It makes sense to contain it in one bp if the goal is consistency, but im curious how you intend to make it consistent
20:19:46 <ediardo> oh yeah,
20:19:55 <robcresswell> unify, as in, make it consistent across panels
20:20:09 <ediardo> http://pasteboard.co/1lqX4Ogp.png
20:20:18 <robcresswell> The issue being that every panel supports different options and even handling, like fuzzy string matching for example for names
20:20:19 <david-lyle> robcresswell: offering the server side filtering facets where available
20:20:21 <ediardo> from the UX perspective, use input-groups
20:20:30 <david-lyle> taking the instance example forward
20:20:32 <ediardo> same filtering by only 1 field
20:20:53 <david-lyle> follow on would be trying to move to magic search
20:21:02 <ediardo> I see that magic search does a good job filtering by several fields and I like how it works, but as fars as i know it does not do any api request...
20:21:05 <david-lyle> but biting all that off at once is too much
20:21:12 <tyr__> searchlight!
20:21:14 * tyr__ ducks
20:21:26 * robcresswell pretends last 2 lines didnt happen
20:22:00 <ediardo> the easiest approach would be do something like the instances table
20:22:08 <ediardo> provide a set of fields and go from there..
20:22:51 <robcresswell> Well, no complaints from me, but I'm very interested in the implementation
20:23:09 <tyr__> no, seriously though...if there is a need to have a consistent search/filtering experience, across APIs that provide different search/filtering capabilities, this could be a lot of very exciting code into Horizon.
20:23:14 * tsufiev cannot deny the merits of Horizon being self-sufficient, but if SearchLight solves it, why solve it the second time?
20:23:36 <tyr__> I'm a big fan of the goal...just sharing robcresswell's concern about the implementation scope
20:23:43 <robcresswell> tsufiev: can't guarantee searchlight
20:24:11 <david-lyle> robcresswell: it's going to be spectacular
20:24:11 <david-lyle> tsufiev: can't name one installation
20:24:24 <robcresswell> Improving the filtering is good though. I'm all for it.
20:24:46 <david-lyle> so either you have to bind the two, or have the fallback of a native horizon solution
20:25:13 <david-lyle> tyr__: if openstack was actually an integrated product, so many things would be much much easier
20:25:33 <david-lyle> but alas the technical direction has been to run in the opposite direction
20:25:50 <tsufiev> robcresswell, yep, I'm not implying "let's wait until SearchLight does that for us"
20:25:54 <david-lyle> so we all have to have shaky dependencies
20:26:19 * david-lyle misses integrated product openstack
20:26:28 <tyr__> david-lyle: like Horizon runs without nova...er...no wait... :)
20:26:34 <tsufiev> personally, I would be glad if there was "SearchLight, but for quotas" :)
20:27:17 <robcresswell> tyr__: Well, thats a Horizon issue, not anything else.
20:27:19 <david-lyle> tyr__: technically horizon should run without nova
20:27:38 <tsufiev> ... and without Keystone
20:27:39 <david-lyle> I think there is a silly legacy check remaining blocking that
20:27:47 <david-lyle> identity is the only true requirement
20:28:20 * david-lyle adds silly check to todo list
20:28:34 <tyr__> ediardo: my .02 is that consistent filtering capabilities is a great goal, but if it starts to get "too useful" it might be nice if it was separable from Horizon, rather than a core component
20:28:41 <robcresswell> Anyway, the blueprint itself is fine. But keep in mind integration with other panels/ services etc. If there ends up being a sensible way to hit searchlight or the APIs (or at least, make the code modular enough that it can be extend in that manner) thats great.
20:29:27 <ediardo> sure, I'm getting my master's degree in apiclients
20:29:33 <r1chardj0n3s> heh
20:29:37 <ediardo> I'll invite u all to the party
20:29:45 <robcresswell> haha
20:30:06 <tyr__> sweet...I'll be asking you about how to invoke a Designate client shortly :)
20:32:14 <robcresswell> Anything else to discuss?
20:32:20 <robcresswell> Otherwise we can end early :)
20:33:02 <robcresswell> Alright, thanks everyone!
20:33:12 <robcresswell> #endmeeting