20:00:48 <david-lyle> #startmeeting horizondrivers
20:00:49 <openstack> Meeting started Wed Sep  2 20:00:48 2015 UTC and is due to finish in 60 minutes.  The chair is david-lyle. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:50 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:52 <openstack> The meeting name has been set to 'horizondrivers'
20:01:02 <robcresswell> o/ I decided to show up for this one
20:01:16 * david-lyle has a feeling there will be 12 different eavesdrop names for this meeting
20:01:54 * david-lyle and robcresswell
20:01:58 <TravT> o/
20:02:14 <tsufiev> o/
20:02:19 <robcresswell> woohoo
20:02:45 <robcresswell> tqtran_ ..?
20:02:53 <tqtran_> o/
20:02:55 <tqtran_> im here
20:03:01 <robcresswell> \o/
20:03:34 <david-lyle> ok
20:03:42 <robcresswell> Agenda: https://wiki.openstack.org/wiki/Meetings/HorizonDrivers#Agenda_for_September_2_2000_UTC
20:03:59 <david-lyle> so this weeks meeting is a choose your own adventure
20:04:19 <ducttape_> I choose network topology
20:04:53 <david-lyle> #startvote todays topic ? discuss potential FFEs for Liberty, discuss the linked agenda
20:04:53 <openstack> Begin voting on: todays topic ? Valid vote options are discuss, potential, FFEs, for, Liberty, discuss, the, linked, agenda.
20:04:55 <openstack> Vote using '#vote OPTION'. Only your last vote counts.
20:05:04 <david-lyle> #endvote
20:05:04 <openstack> Voted on "todays topic ?" Results are
20:05:07 <david-lyle> bah
20:05:17 <robcresswell> go for FFEs, makes more sense
20:05:18 <hurgleburgler> I was gonna vote for the
20:05:30 <TravT> fastest vite ever
20:05:34 <david-lyle> I didn't put commas in WTH
20:05:55 <david-lyle> vote informally
20:06:05 <tqtran_> ffe is fine
20:06:21 <tqtran_> although, wanted to point out https://blueprints.launchpad.net/horizon/+spec/transfer-table-clone-feature is already done with
20:06:45 <david-lyle> tqtran_: I noticed
20:06:53 <david-lyle> I was cleaning up bps yesterday
20:07:31 <david-lyle> #link https://launchpad.net/horizon/+milestone/liberty-3
20:08:04 <david-lyle> and
20:08:07 <david-lyle> #link https://launchpad.net/horizon/+milestone/liberty-rc1
20:08:13 <jwy> i still have 2 patches for https://blueprints.launchpad.net/horizon/+spec/import-images out for review, which is targeted for liberty-3. would that be able to make it in?
20:08:20 <david-lyle> that's what we are officially tracking right now
20:09:30 <david-lyle> jwy: you need two core reviewers to look at it
20:09:40 <david-lyle> any volunteers?
20:09:54 <TravT> trying to read...
20:09:57 <david-lyle> I have a list of sahara and trove that I'm slowly working through
20:10:12 <robcresswell> its 9pm here, can I respectfully decline?
20:10:16 <robcresswell> :p
20:10:23 <david-lyle> robcresswell: not today
20:10:25 <TravT> david-lyle: i can review the import images patches, but not before tomorrow.
20:10:38 <david-lyle> at this point we're looking at FFEs
20:10:56 <david-lyle> gate is ~15 hours deep
20:11:02 <david-lyle> and we'll tag tomorrow
20:11:41 <david-lyle> so I think we could handle 5-6 FFEs if people sign up to shepherd them through
20:12:18 <david-lyle> so the question is what's high importance and who has bandwidth
20:12:37 <david-lyle> I think curvature is almost there, but I know there were some questions about usability
20:12:48 <david-lyle> jwy: will circle back
20:13:02 <jwy> david-lyle: thanks
20:13:13 <david-lyle> trying to get a better estimate on load
20:13:20 <tqtran_> theres a also a lot of code to look through, wondering if it make sense to package that externally, otherwise it would have to pass all of our linting rules
20:13:37 <david-lyle> tqtran_: which part?
20:13:43 <david-lyle> the curvature?
20:13:47 <tqtran_> the curvature code
20:14:00 <robcresswell> Okay, so there are currently 1 at L-3, and 3 at RC1. One of those at RC1 is just that last docs patch, its pretty minor.
20:14:15 <robcresswell> I'll look at Curvature. I know it already anyway.
20:14:23 <david-lyle> I'll have more plugin docs for review in RC-1 as well
20:14:34 <david-lyle> but docs are easier
20:14:36 <robcresswell> But need someone non-cisco to look over too
20:15:00 <david-lyle> tqtran_: curvature is not currently covered by the linting checks?
20:15:17 <tqtran_> it is not, we made an exception for legacy js files
20:15:54 <david-lyle> #link https://blueprints.launchpad.net/horizon/+spec/curvature-network-topology
20:16:00 <david-lyle> for the viewers at home
20:16:04 <ducttape_> so then let's keep that, and follow on with curvature lint cleanup ?
20:16:49 <ducttape_> or is lint cleanup more important?
20:16:52 <tsufiev> ducttape_, there is a patch to angularize curvature code, might be worth to do this work there
20:17:20 <david-lyle> looks like some of the usability/visualization issues have been addressed
20:17:24 * david-lyle has not retried
20:17:33 <TravT> my concern's on curvature were more usability... I haven't tried latest patches
20:17:44 <TravT> but i didn't go through all the code again.
20:17:46 <tqtran_> tsufiev: can you provide a link to this?
20:17:47 <david-lyle> this has been a long time coming, I think if it's ready we should put it in
20:18:17 <david-lyle> looks like it's already targeted
20:18:25 <tsufiev> tqtran_, https://review.openstack.org/#/c/199063/ based on some older patchset, some work to do there
20:18:58 <david-lyle> reminder, targeting is not a guarantee that it will get in, just if the code is ready
20:19:00 <tqtran_> oh wow, thats a lot farther than i thought, cool
20:19:02 <ducttape_> the curvature stuff does not always create a great experience for what items are drawn where on first load.... but if you have a large / complex network it makes a HUGE improvement over what we have today
20:19:03 <TravT> david-lyle: should it go in as a full replacement, or configurable alternative?
20:19:31 <david-lyle> so there was a suggestion to have both views
20:19:48 <david-lyle> at one point, do others feel there's value in that
20:20:23 <robcresswell> If Curvature is available, would the old one be used still? I guess thats the value question.
20:20:26 <david-lyle> I think that's too much to tackle at this late time, even if people like the idea
20:20:50 <ducttape_> there's value if not everyone is ok with the change.  it's like launch instance, where it might take a while to get people moved over to using it
20:20:50 <david-lyle> I really think it should be an all or none
20:20:56 <tsufiev> +1 for having the both views
20:21:26 <TravT> the downside is that then we have two paths of code we're dealing with.
20:21:32 <robcresswell> That will really hold the patch up...
20:21:59 <robcresswell> I'm for a clean swap, but if its gonna annoy the deployers, then we should use both.
20:22:21 <tqtran_> my vote is for clean swap as well
20:22:21 <ducttape_> I'm one voice, but I would switch to new topology before I switched to new launch instance
20:22:53 <tsufiev> ducttape_, +1 :)
20:23:02 <jwy> is there any functionality from the old one that's not in the new one?
20:23:05 <ducttape_> we might actually carry the topo patch locally and deploy it that way, if it gets held up
20:23:18 <robcresswell> Yeah, so... lets just go for Curvature
20:23:31 <david-lyle> I think if it's ready, we should merge it
20:23:56 <tqtran_> my only concern is, once we merge it, thats more technical debt we have to tackle
20:23:57 <ducttape_> jwy:   there are some anoyances with the UX of the new topo stuff, but features are all the same
20:23:57 <robcresswell> We've got a week or so to make sure its right on the mark. And brad is back from moving house, he'll be babysitting the patch through.
20:24:23 <ducttape_> what is the effort to get lint cleanup on it?
20:25:03 <tqtran_> ducttape_: not sure, we'll have to enable it on legacy and find out
20:25:13 <tsufiev> ducttape_, personally I don't see much value in polishing some of the code it uses, especially pop-ups - they'd better to be rewritten completely
20:25:16 <tqtran_> right now, it all shows up as warnings
20:25:39 <tqtran_> my vote would be to improve tsufiev
20:25:43 <tsufiev> I mean the legacy pop-up code that was incorporated into network topology from previous view
20:25:48 <tqtran_> patch's and work from there
20:25:56 <david-lyle> what's wrong with tsufiev?
20:26:04 <tsufiev> I'm totally fine )))
20:26:07 <robcresswell> haha
20:26:08 <tqtran_> nothing lol, i pressed return by accident
20:26:11 <ducttape_> it needs more sizzle ;)
20:26:13 <david-lyle> :)
20:26:18 <tqtran_> hahaha
20:26:18 <lhcheng> hah
20:26:22 <tqtran_> sorry tsufiev!
20:26:35 <tsufiev> tqtran_, no worries, it was funny )
20:26:36 <david-lyle> oh, let's leave it out of the linter
20:26:47 <tqtran_> hahaha ><
20:27:06 <david-lyle> if the bouncy d3 rendering doesn't drive me crazy, I'd like to see it merge
20:27:30 <david-lyle> ok what's next?
20:27:32 <robcresswell> `npm run lint` on the curvature patch has about 6 warnings... thats fairly minor
20:27:55 <ducttape_> maybe we can get those cleaned up and that would be sufficient cleanup ?
20:28:12 <tsufiev> tqtran_, I tend to agree with your decision, not because it's my patch, but also because that way you'll some time by not doing the stuff I already did when Angularized it
20:28:24 <tsufiev> *save some time
20:28:37 <robcresswell> The other 2 bps floating around are jwy one from earlier, and this Sahara one https://blueprints.launchpad.net/horizon/+spec/allow-editing-of-job-binaries
20:28:43 <TravT> I'd like to see FFE for both users and images NG table work.
20:28:49 <tqtran_> plus, the break up of logical parts will make it much easier to review and scrutinize
20:29:39 * robcresswell always forgets to add link :(
20:29:56 <david-lyle> my preference would be to send in the curvature code without lint fixes unless they are trivial
20:30:10 <david-lyle> sounds like there is work to rework it for M already in progress
20:30:34 <david-lyle> ok, I have the job-binaries one
20:30:40 <david-lyle> I have a -1 on it
20:30:48 <david-lyle> if others want to review go for it
20:31:06 <david-lyle> I think trove clustering should be in too
20:31:09 <david-lyle> over a year old
20:31:38 <robcresswell> Is that the 1700 line patch?
20:32:14 <david-lyle> yup
20:32:43 <robcresswell> siiigh fine I'll help review it to
20:32:50 * robcresswell sulks
20:32:53 <david-lyle> just needs one core
20:32:59 <david-lyle> it's in contrib now
20:33:19 <tqtran_> whats the link to the patch?
20:33:23 <david-lyle> lhcheng has started looking
20:33:27 <david-lyle> I have too
20:33:35 <tqtran_> ok, then i'll leave it to you guys lol
20:33:41 <david-lyle> https://review.openstack.org/#/c/110411/
20:34:12 <lhcheng> david-lyle: just did a first pass, had to take a break in between
20:34:28 <lhcheng> 1700 line is a lot
20:34:34 <david-lyle> lhcheng: understood, I've taken many breaks :)
20:34:38 <david-lyle> maybe mostly breaks
20:35:57 <david-lyle> angular docs bp is fine
20:36:07 <tqtran_> so... going back to what TravT mentioned, i'd also like to see FFE on users and images ng work
20:36:13 <david-lyle> ok now NG users and images was the other proposal
20:37:27 <TravT> there are a number of patches on images.
20:37:39 <TravT> undergoing review.
20:37:43 <TravT> example: https://review.openstack.org/#/c/217422/
20:38:23 <david-lyle> I'd like to take input before I go. I tend to just jump in and stomp on the conversation, so I'll hold back for a minute
20:38:42 * TravT prepares for the beating he is about to receive
20:38:56 <robcresswell> Would the FFE be in anticipation of making these live at Liberty release?
20:39:19 <TravT> you mean enabled by default?
20:39:44 <robcresswell> possibly. But really I meant, fully funtional
20:39:59 <TravT> i think for me, we got a lot of requests in kilo for people wanting to know how to build angular plugins
20:40:11 <TravT> this will continue
20:40:22 <ducttape_> +10
20:40:27 <TravT> and the further we push the effort forward, the better we can answer those questions
20:40:41 <robcresswell> Rather than buried; if they;re intending to be usable in the next 2 weeks, then it should be FFE work. If it won't be finished in that timescale, we should freeze it for now? I thought that was the point of a freeze.
20:40:57 <robcresswell> Sure, makes sense
20:41:29 <robcresswell> I'm just wondering if adding more new code to something we don't intend to have prod ready, is sensible for an FFE?
20:41:44 <robcresswell> But if it will be ready, then yeah, lets do it.
20:42:08 <TravT> depends on your definition of ready.
20:42:21 <ducttape_> if it's not prod ready, I hope it's still in gerrit
20:42:39 <vgridnev> Hey, folks! What about one more FFE for Sahara staff on https://blueprints.launchpad.net/horizon/+spec/sahara-node-processes-tab ? It just an simple re-org, and actual patch-set is ready on review
20:42:46 <TravT> curvature seems to mostly get a "looks cool" let's ship it level of review.
20:43:16 <ducttape_> TravT - current topo has a bug, and new one does not.  so feature wise it is more than just sizzle
20:43:29 <ducttape_> at least for us
20:43:41 <tqtran_> i think functionality-wise, the users panel is all there. but the code quality is not quite where we want it. there are a few patterns we think we can abstract out
20:44:18 <tqtran_> and since the images panel is base off the users panel, i would say in 2 weeks time, functionality-wise, it can be pretty solid and usable as well
20:44:36 <ducttape_> but still laking good generalized patterns ?
20:44:49 <lhcheng> if it is not quite ready, why do we want people to follow that pattern?
20:44:56 <tqtran_> we have a pretty good general pattern
20:45:04 <TravT> I do believe there is opportunity to abstract some reusable code with other ng tables doing batch delete, which will ultimately reduce boilerplate looking code and make it a bit more configuration oriented. However, I also believe you need to get a couple concrete, working examples in order to find a pattern.
20:45:54 <ducttape_> so would it make sense to have test cases driving some core classes - to flush this stuff out ?
20:46:14 <tqtran_> https://drive.google.com/file/d/0B5nlaOV3OEj5aVpRaC1sdlgtQ28/view?usp=sharing
20:46:19 <ducttape_> and not try to implement users or something that people might pick up and use?
20:46:25 <tqtran_> heres the general pattern we are using for table actions
20:46:42 <tqtran_> all the ones with jigsaw puzzle are extensible
20:48:17 <tqtran_> hm... that one isnt working for me: https://drive.google.com/file/d/0B5nlaOV3OEj5aVpRaC1sdlgtQ28/view
20:48:27 <ducttape_> I don't think there is consensus on this stuff, could be from lack of understanding or reviewing the state of things
20:48:45 <tqtran_> basically, i took what we talked about in the ML and applied it to make it more extensible and pluggable
20:49:04 * ducttape_ can see both links
20:49:21 <tqtran_> ducttape_: right, i think thats part of the problem. i'm doing my best to try and engage people
20:49:22 <TravT> i don't believe we'll have it all perfect in two weeks.
20:49:45 <TravT> but i think we can improve upon what is there with a few patches.
20:50:17 <ducttape_> would it make sense for those less familiar (like lhcheng) to wait for those patches?
20:50:28 <ducttape_> or review what is there now ?
20:50:38 <TravT> anybody can review at any time
20:50:44 * ducttape_ likes volunteering lhcheng for work
20:50:58 * lhcheng got volun-toss
20:51:08 <david-lyle> LOL
20:51:50 <lhcheng> david-lyle: got a DOA patch needed for L too: https://review.openstack.org/#/c/219041/
20:52:02 <lhcheng> keystone patch being merge atm
20:52:21 <lhcheng> after waiting 10 hrs, the patch needs a recheck..
20:52:29 <david-lyle> lhcheng: ok, that one's on my radar
20:52:38 <TravT> i hate to change subjects, but also want to ask lhcheng: david-lyle: how are the domain patches going?
20:52:40 <david-lyle> ok, I'm torn. I would absolutely like to see progress made on the angularized tables
20:52:48 <david-lyle> so funny  you should ask
20:53:03 <david-lyle> we're going to hold them until post Liberty
20:53:08 * ducttape_ thinks there is nothing funny for domains
20:53:21 <david-lyle> because they won't be useful to horizon in Liberty
20:53:45 <lhcheng> TravT 1 of the main DOA patch is ready, but not merged.  It doesn't make sense to merge without the 2nd half from Dan.
20:53:52 <david-lyle> and merge them in 1.5.0 when the stable branch is cut
20:54:20 <david-lyle> the second patch needs more, but even if both are ready, at this late point, more risk to destabilize
20:54:29 <lhcheng> ++
20:54:44 <david-lyle> and that's the goal of the FF
20:54:56 <lhcheng> once M opens up, we can merge it, so Dan can start working on the horizon side early.
20:55:10 <david-lyle> get ready stuff in, while maintaining stability
20:55:43 <david-lyle> the NG items have the benefit that if they are disabled then they aren't very destabilizing, but again, is there value in that
20:56:26 <david-lyle> and I fully understand getting things in particular releases means more to some than others, we all have day jobs
20:57:03 <lhcheng> same logic why we haven't merge Dave's patch on domain scoping.
20:57:59 <david-lyle> I'm tempted to give the 1-2 NG items a FFE and see where it gets to, but if it's not ready not merge it
20:58:12 <TravT> they are broken down into smaller chunks.
20:58:17 <david-lyle> because I think it's existence has value if it's a reusable pattern
20:58:22 <TravT> so we could look at pieces of it at a time.
20:59:28 <lhcheng> would the work kfox have, is a better model for NG plugin? Since it is really an external plugin which operator would do..
20:59:47 * lhcheng not sure what
20:59:48 <robcresswell> So... lets give the ng work FFE then? It seems enough people are happy with it being allowed to proceed, see where it gets to, then merge or not.
20:59:57 * lhcheng not sure what the state right now
21:00:18 <david-lyle> lhcheng: true, builds on the infrastructure developed as a by product of the NG content bps
21:00:26 <tqtran_> kfox's work is based on what we are doing, its really just a simple extension of our current plugin model
21:00:30 <robcresswell> Just trying to judge community decision
21:00:32 <robcresswell> :)
21:00:41 <lhcheng> and focus on making that working on L.
21:01:02 <lhcheng> tqtran_: ah got it
21:01:08 <TravT> maybe we could just avoid this whole hassle and go to more frequent releases
21:01:33 <david-lyle> TravT: that's a possibility, but brings it's own complications
21:02:06 <tsufiev> iterations, agile, scrum - all that stuff
21:02:07 <david-lyle> ok out of time, what's the consensus
21:02:44 <robcresswell> I say FFE. I don't have strong views either way, and its disabled by default so I suppose unlikely to break.
21:03:08 <robcresswell> More patterns is good and clearly something desired by community, right?
21:03:42 <david-lyle> alright, I will grant FFEs, but I want the core group wider on those
21:03:49 <david-lyle> robcresswell: do you have time?
21:04:05 <robcresswell> Yeah, sleep is for the weak
21:04:19 <robcresswell> I kid, I have time.
21:04:20 <david-lyle> we have the better part of 3 weeks
21:04:34 <david-lyle> but it better not be landing at the end
21:04:40 <david-lyle> I will look to
21:05:13 <david-lyle> Thanks everyone
21:05:18 <david-lyle> #endmeeting