20:00:42 <robcresswell> #startmeeting horizon
20:00:43 <openstack> Meeting started Wed Aug  3 20:00:42 2016 UTC and is due to finish in 60 minutes.  The chair is robcresswell. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:44 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:47 <openstack> The meeting name has been set to 'horizon'
20:00:48 <robcresswell> o/
20:00:51 <rdopiera> o/
20:01:13 <yohoffman> \o
20:01:21 <ezpz_> o/
20:01:32 <robcresswell> Hi everyone
20:01:36 <Marcellin_> o/
20:01:51 * robcresswell slowly finds agenda
20:02:01 <david-lyle> o/
20:02:17 <robcresswell> Okay, lets get started
20:02:26 <robcresswell> We're 3 weeks from FF
20:02:28 <TravT> o/
20:02:29 <tsufiev> o/
20:02:40 <robcresswell> So cores need to be reviewing reviewing reviewing
20:03:00 <robcresswell> And we need people to be on the look out for bugs, finalising their features.
20:03:41 <robcresswell> I'll be running around getting django up to date and reviewing priority items
20:04:00 <robcresswell> Please focus reviews on:
20:04:03 <robcresswell> #link https://review.openstack.org/#/q/starredby:rob-cresswell+status:open
20:04:31 <robcresswell> I've left that in the Horizon IRC room topic too.
20:04:52 <robcresswell> Any questions about FF/ release?
20:05:19 <robcresswell> The release schedule, FYI, is here: http://releases.openstack.org/newton/schedule.html
20:05:56 <robcresswell> Lets move on to the agenda
20:06:01 <robcresswell> #topic Agenda
20:06:08 <robcresswell> #link https://wiki.openstack.org/wiki/Meetings/Horizon#Agenda_for_2016-08-03
20:06:19 <robcresswell> Couple of items
20:06:27 <robcresswell> Marcellin_: o/
20:06:48 <robcresswell> So Marcellin_ reached out last week asking for any features that need extra input
20:06:55 <r1chardj0n3s> o/
20:07:19 <robcresswell> If there are any small bps that could be finished off, or larger bps for O-1, please speaker to Marcellin_ about contributing there.
20:07:23 <robcresswell> speak*
20:07:39 <tsufiev> robcresswell, 0-1?
20:07:49 <robcresswell> next cycle
20:07:55 <r1chardj0n3s> tsufiev: get a better font ;-)
20:08:11 * tsufiev has quotas in mind
20:08:18 <robcresswell> They're looking to contribute to bps, which obviously may be tough with FF so soon, hence the reference to next cycle :)
20:09:02 <robcresswell> Marcellin_: Anything to add?
20:09:07 <tsufiev> lol, I was thinking about some score in a soccer game :)
20:09:19 <r1chardj0n3s> :-)
20:09:24 <ezpz> I work with Marcellin_ and would be willing to help as well
20:09:42 <robcresswell> Oh excellent
20:10:08 <robcresswell> Hopefully someone will be able to throw some useful bps your way soon :)
20:10:12 <robcresswell> Lets move on for now
20:10:21 <robcresswell> #topic UI autotests
20:10:29 <robcresswell> schipiga: o/
20:10:30 <schipiga> Hi, all! I would like to propose redesigned horizon autotests https://github.com/sergeychipiga/horizon_autotests, based on steps-architecture. And according to this proposition I want to prepare gerrit patch with these tests.
20:10:55 <schipiga> Current advantages, that I see: strong architecture, reusable code (not copy-paste!), stable microkernel (https://github.com/sergeychipiga/pom), parallel tests launching, easy tests coding including tests for angular forms (right now there are over 40 tests for angular forms: containers and instances).
20:11:16 <schipiga> And because I put the soul to these tests, I'm going to provide readable documentation for these tests and architecture, to provide low entry threshold for new developers.
20:11:36 <robcresswell> 40 tests??
20:11:40 <robcresswell> That was fast
20:11:52 <schipiga> 40 for new angular forms
20:11:55 <r1chardj0n3s> +1
20:12:17 <schipiga> but all 121 tests for almost all UI components
20:12:31 <schipiga> images, volumes, instances, users, etc
20:12:41 <tsufiev> how long does it take to run them in a single thread on a typical Jenkins node?
20:12:43 <robcresswell> So, what is this replacing? Integration? Or is a functional or unit level?
20:12:49 <robcresswell> Sorry, I've not yet looked into the repo
20:13:43 <tsufiev> robcresswell, afaik schipiga is aiming to replace existing integration tests
20:14:38 <schipiga> tsufiev in our jenkins inside company, On average each test requires 35 seconds vs 46 seconds in current upstream version
20:15:22 <robcresswell> This sounds like a positive move
20:15:38 <tsufiev> which gives 46 seconds x 121 = 6677 seconds = 92 minutes
20:15:41 <robcresswell> Can we get it blueprinted and a POC proposed inside Horizon?
20:15:41 <schipiga> and right now they support parallel launching with pytest-xdist
20:16:11 <TravT> sounds pretty wonderful, what's the catch?
20:16:18 <tsufiev> yep, PoC of how well Jenkins slaves handle >=2 threads of execution would be nice
20:16:36 <tsufiev> more precisely, how Devstack handles it :)
20:16:39 <robcresswell> tsufiev: On the plus side, even two threads would be 46 minutes
20:16:56 <tsufiev> not exactly, robcresswell
20:17:12 <schipiga> TravT, not in tests, but infra must have nice hardware to provide over 2 test threads simultaneously
20:17:17 <tsufiev> they can start interfering each other and producing intermittent failures if resources are not enough
20:17:18 <robcresswell> Yeah, I should have said up-to
20:18:47 <schipiga> robcresswell, tsufiev, yep, there may be problems with openstack performance in parallel launching. But in single-threaded mode, they work fine already (they are used inside company already)
20:19:45 <robcresswell> This is all theory regardless.
20:20:01 <robcresswell> schipiga: It sounds good, and certainly something we would be interested in
20:20:15 <robcresswell> A POC in Horizon would be a good next step
20:20:26 <robcresswell> So we can see it functioning in the gate
20:20:37 <schipiga> robcresswell, ok, I'm ready to integrate these tests to horizon upstream
20:21:30 <robcresswell> Sounds good
20:21:41 <robcresswell> #topic Open Discussion
20:21:51 <robcresswell> There are no more agenda items, so we have an open floor
20:22:10 <robcresswell> Any questions, discussion while we have a large group here, please go ahead :)
20:22:49 <robcresswell> Also I'll be pinging people fairly often for reviews over the next month to hit release. You've been warned :)
20:23:26 <r1chardj0n3s> ui-router? ui-router! https://review.openstack.org/#/c/350523/
20:24:11 <r1chardj0n3s> (note to self: I need to update my bower-to-xstatic script so you don't need to run xstatic-release on the result ;-)
20:25:27 <robcresswell> I could've sworn someone argued strongly against adding ui-router a few months back...
20:25:31 * robcresswell ducks
20:26:36 <robcresswell> r1chardj0n3s: Is the repo under openstack control or yours?
20:26:49 <robcresswell> I still have a couple for schema form that need to be put under the OS namespace
20:27:25 <robcresswell> Is ui-router being used exclusively for this use case? Are there others where it would be useful?
20:27:46 <r1chardj0n3s> robcresswell: yeah yeah :-P the repo is currently local, I need to get it into openstack
20:28:10 <r1chardj0n3s> it'll be helpful in some work that tyr_/TravT are doing (designate IIRC)
20:29:24 <robcresswell> TravT, tyr_: thoughts?
20:29:45 <TravT> well, i originally proposed we consider ui router
20:30:27 <TravT> so, i'm not opposed to it conceptually...
20:30:29 <r1chardj0n3s> yep, also, I had very tasty humble pie a few days ago, fyi
20:30:33 <TravT> just need to look at it
20:30:53 <robcresswell> r1chardj0n3s: hahaa
20:31:27 <TravT> i don't like bringing something like this in late in the release, but i also think sooner is better overall
20:31:59 <tyr_> I personally haven't hit the need for ui-router. I don't have particular comments. If it is a more powerful way to do routing, that seems good on paper.
20:32:03 <r1chardj0n3s> ui-router has some warts, like what I mentioned last night about paths
20:32:24 <r1chardj0n3s> its documentation is also so-so - a lot of head scratching :/
20:32:27 <tyr_> you want to bring this in for Newton?
20:32:39 <r1chardj0n3s> (these are reasons I was hesitant the first time it was brought up)
20:33:02 <r1chardj0n3s> I need to complete the WIP for swift ui before I'd be fully comfortable pushing it in as there's some open questions
20:33:11 <robcresswell> I guess the blunt question would be whether you have the time to introduce it and bugfix with pycon also soon
20:33:20 <r1chardj0n3s> also if anyone has experience with ui-router I'd appreciate a glance at the patch
20:33:29 <robcresswell> I don't have any particular issue with a new requirement (packagers might...)
20:33:45 <r1chardj0n3s> robcresswell: pycon is really just the next week and half, so outside of that I'm confident I have time
20:33:49 <robcresswell> So may be worth speaking to mrunge or zigo
20:33:59 <robcresswell> r1chardj0n3s: ah, great
20:34:03 <TravT> does this affect out of tree plugins?
20:34:09 <TravT> will we have to go migrate them?
20:34:23 <r1chardj0n3s> AFAICT it doesn't affect anything
20:34:29 <TravT> sorry that i haven't actually tried the patch
20:34:35 <TravT> doing too many things at once
20:34:37 <r1chardj0n3s> it's an additional method of routing, not a replacement. it doesn't make ngRoute not work
20:34:58 <TravT> i suppose mostly then it is a testing exercise
20:35:14 <r1chardj0n3s> well, it's also a sanity check that what I'm doing is good and correct :-)
20:35:56 <r1chardj0n3s> I'm not that familiar with ui-router, since I discarded it as a bit too rough a few years ago when I first looked at i
20:36:07 <r1chardj0n3s> it's still a bit rough, just not *as* rough
20:36:25 <r1chardj0n3s> but it's also the only solution for nested routing that I'm aware of
20:37:17 <robcresswell> r1chardj0n3s: What about versioning? Do we need 1.5 for it? Or is the current 1.4 fine?
20:37:35 <r1chardj0n3s> you mean angular 1.5?
20:37:40 <robcresswell> Yup
20:37:53 <robcresswell> Just wondered how backwards compatible the lib is
20:38:04 <r1chardj0n3s> I'm using 1.4.10 at the moment and it's working
20:39:28 <robcresswell> cool, just checking
20:39:47 <tyr_> r1chardj0n3s: at first blush, nested routing seems like a complex way to solve a problem where we would normally just use a component like a TabNavigator or Panel.
20:40:11 <robcresswell> I've no issue with this. Lets review the POC and see if it feels like the correct solution and the code is clean etc.
20:40:13 <tyr_> probably need to play with swift UI to better understand the motivation.
20:41:30 <robcresswell> r1chardj0n3s: I'll take a look at the patch probably late this week/ early next week
20:42:21 <robcresswell> I don't know if anyone saw the mailer post, but 1.5.x is fairly ready, just need to double check tests. functionally it was fine though. ALl the images content worked.
20:42:57 <r1chardj0n3s> tyr_: that's somewhat true, but the swift ui is a special UI snowflake. Perhaps it shouldn't be...
20:43:29 <tyr_> if there is tension between landing schema form, and landing ui-router, I have a much stronger desire to see schema form land.
20:43:38 <r1chardj0n3s> absolutely
20:44:47 <robcresswell> r1chardj0n3s: Does the state charing problem with the top level containers and the objects go away if you design the UI around a single table?
20:44:58 <TravT> why would there be tension between those two things?
20:45:43 <r1chardj0n3s> robcresswell: quite possibly. The containers side list complicates things, yes.
20:46:02 <r1chardj0n3s> TravT: I can't see any myself, but in case there ever was, I'm on schemaform's side too :-)
20:46:22 <robcresswell> I don't know why they would be competing, think we all missed something
20:46:26 <tyr_> r1chardj0n3s: are you hoping for ui-rotuer in Newton?
20:46:35 <TravT> let's get schemaform in.
20:46:50 <r1chardj0n3s> I need to to a bunch more work on the ui-router patch - I just put it up to get any feedback on my use of ui-router from any experts in the team :-)
20:46:59 <r1chardj0n3s> tyr_: that'd be nice yes
20:47:52 <TravT> sounds good r1chardj0n3s
20:48:09 <tyr_> ok, my point is just that investigating ui-router will have a minor distracting effect from landing in-flight Newton work.
20:48:18 <r1chardj0n3s> yup
20:48:31 <r1chardj0n3s> happy for it to not be that distraction
20:50:09 <robcresswell> Given the reaction, it doesn't seem like we'll be able to get the input on it to make an informed decision
20:50:20 <r1chardj0n3s> yup, that's my feel
20:50:24 <r1chardj0n3s> I won't push for N
20:50:33 <robcresswell> Leave the POC up, and we'll see if we can get some input over the next week. If not, we'll defer to O.
20:50:37 <r1chardj0n3s> yep
20:50:41 <robcresswell> Cool
20:51:04 <TravT> robcresswell re: 1.5.x, i guess since we waited until the end of release for 1.4.x, that we could push 1.5.x and try out over the next few weeks.
20:51:56 <TravT> i have no idea what deployers opinion would be on that
20:52:10 <TravT> err... packagers
20:52:37 <robcresswell> packagers would probably prefer 1.5 since its more recent. Well, zigo would, since he packages edge
20:52:48 * robcresswell forgets debians track names
20:52:55 <tsufiev> sid
20:53:01 <robcresswell> ta, tsufiev :)
20:53:14 <tsufiev> bleeding edge, i would say ;)
20:53:15 <robcresswell> Anyway, 5 minutes left: any further questions, points to raise?
20:54:21 <robcresswell> I think we can call it there. Thanks everyone!
20:54:29 <robcresswell> #endmeeting