22:02:39 #startmeeting zuul 22:02:40 Meeting started Mon Dec 18 22:02:39 2017 UTC and is due to finish in 60 minutes. The chair is jeblair. Information about MeetBot at http://wiki.debian.org/MeetBot. 22:02:41 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 22:02:43 The meeting name has been set to 'zuul' 22:02:44 o/ 22:02:55 #link agenda https://wiki.openstack.org/wiki/Meetings/Zuul 22:03:02 #link previous meeting http://eavesdrop.openstack.org/meetings/zuul/2017/zuul.2017-12-11-22.01.html 22:03:05 #topic roadmap 22:03:35 we've cleared a *lot* of stuff off the road map! 22:03:45 #link roadmap changes: https://etherpad.openstack.org/p/FFbJanDCQ7 22:03:46 Indeed 22:03:58 i just went over the list, and that's all the 3.0 blockers sorted by status 22:04:13 https://storyboard.openstack.org/#!/search?tags=zuulv3.0 22:04:26 Lot 'o green 22:04:41 * mordred waves 22:04:44 since the last meeting we've merged nodepool quota, zuul-web, the git driver, the finger gateway, the private api key url fix, and bumped the gitpython dep! 22:05:04 okay i lied about the git driver 22:05:17 it's been approved but ran into the sphinx issues 22:05:20 it's as good as merged though 22:05:24 https://storyboard.openstack.org/#!/story/2001336 is done then? 22:05:39 Oh right 22:05:41 lagfail 22:06:07 so YAAAY! thanks everyone for doing those! 22:06:23 o/ 22:06:25 nice work :-) 22:06:46 jeblair: fwiw, sphinx fix has landed, rechecking the git driver patch 22:06:55 and thanks to the the reviewers, and folks that have stuff in progress -- which is most of the rest of the list! :) 22:07:13 does it mean zuul-web now consumes GH POST ? 22:07:31 I don't think that part is quite done. 22:07:43 https://storyboard.openstack.org/#!/story/2001327 22:07:55 rcarrillocruz: oops missed that on the etherpad list. i think that one is in progress? 22:08:12 ack 22:08:18 some combination of tobiash/jlk are working on that, i think? 22:08:33 Mostly tobiash at this point, I'm distracted elsewhere 22:08:50 My understanding is that tobiash is making the GHE tech preview of apps work. 22:08:50 cool 22:08:50 yeah, someone had been rebasing, just wondered i missed that getting merged 22:09:21 but could be also working on the ingestion in zuul-web 22:09:28 it may be worth a checkin with tobiash to see if that's still on his list or if it could use another volunteer to push it along 22:10:00 mordred: should we talk about the javascript patches for a sec? 22:10:20 ++ - it might also be worthwhile to either rebase that patch on top of tobiash's patches for ghe apps or vice versa - but certainly in coordination with tobiash 22:10:22 jeblair: yup 22:10:22 i haven't dived in to review them yet -- will probably be tomorrow at this point for me 22:10:51 so - I was doing to write an email about what reviewers might want to look at - but I'll just do that here :) 22:11:29 the first and most important thing is - do the results still work - just like storyboard, the patches add a 'draft' publication that points to the live zuulv3 endpoint 22:11:39 that's so cool i love it 22:11:43 so: http://logs.openstack.org/38/487538/15/check/build-javascript-content/3432950/npm/html/status.html?source_url=http://zuulv3.openstack.org 22:12:09 if you switch views, you have to stick the source_url on the end again, yeah? 22:12:17 the builds and jobs links from the status page do not curently work in the draft setup - I'll get that fixed soon 22:12:22 jeblair: yes - that's where I was getting with that 22:12:50 that's.. kind of amazing 22:13:19 builds don't work right in the first patch ... 22:13:27 but if you go down the stack a little ways, http://logs.openstack.org/98/528298/10/check/build-javascript-content/2e5b6ac/npm/html/builds.html?source_url=http:%2F%2Fzuulv3.openstack.org 22:13:34 Note that it makes me also want to have a static "good example" json that we stick in there that can be used so that they don't have to do the whole dance with zuulv3.openstack.org 22:13:35 http://logs.openstack.org/98/528298/10/check/build-javascript-content/2e5b6ac/npm/html/jobs.html?source_url=http:%2F%2Fzuulv3.openstack.org 22:13:37 ah goot to know :) 22:13:40 SpamapS: we have one! 22:13:45 SpamapS: (lucky for you) 22:13:51 mordred: did the old status vs new status stuff get untangled in the stack (that was the one thing I ran into when I briefly poked at things) 22:14:04 clarkb: old status / new status? 22:14:10 clarkb: can you define "old status vs new status"? 22:14:23 jeblair: mordred was making changes to make all the old status stuff keep working, but it should just be deleted imo and reduce the review overhead 22:14:36 clarkb: yes- it's all collapsed into one 22:14:49 ++ 22:15:00 SpamapS: your question leads to the next thing I'd suggest people review ... which is the README file in web/README.rst 22:15:18 I tried to write it so that it has enough information for people to casually touch it or for people who want to go deeper 22:15:58 I *highly* recommend trying out "npm run start:status" - as that launches the status page in a local dev webserver - and is plubmed in to trigger hot refreshes of the app when you make local code chagnes 22:16:21 (and for me is one of the things that's nice enough to make me value learning the new tools) 22:16:42 mordred: my first reaction was "I'm glad pip doesn't run a web server" 22:17:06 clarkb: yah. :) 22:17:28 clarkb: 'npm run build' will justmake the static files as if they were for production and then put them into zuul/web/static 22:18:08 other things to mention - the build produces 'map' files, so that exceptions thrown in browser can be debugged with pre-minified source code (which is nice) 22:18:31 the stack also sets up transpiling using a javascript tool called "babel" (not the same as the python one with the same name) 22:18:39 mordred: are you still considering move the JS / web bits in different repo (eg: zuul-dashboard)? 22:18:51 that allows us to write modern javascript without having to worry about old browsers 22:19:06 mordred: so its compiling js to js 22:19:10 just different versions of browsres 22:19:22 pabelanger: not today - that's a deeper conversation that has pros and cons, but the stuff is done right now assuming all-in-one-repo 22:19:29 clarkb: yah 22:20:02 ack 22:20:10 oh - that brings up the largest gotcha on my TDL for this code - which is that we have a zuul unittest that assumes the existence of the html/js files (it checks that streaming is working) 22:20:35 mordred: which test? 22:20:46 I'm assuming it's going to take a few days for folks to review - so I should have a solution to that before it's time to land (same thing with the status/builds/jobs links) 22:21:09 jeblair: https://review.openstack.org/#/c/487538/15/tests/unit/test_streaming.py 22:21:10 couldn't you just test the python bits directly? 22:21:38 yah - that's what I'm thinking - rewrite it to just hit the websocket API 22:21:41 mordred: i thought that just opened a websocket connection? 22:22:10 jeblair: it becomes rather unhappy if the html files aren't there - it's next on my list of things to debug properly 22:22:55 mordred: ok cool. i think the *intention* is that it's just testing websocket, so i think a fix to that that ignores any static/js deps that have leaked in sounds reasonable 22:23:06 I *think* it's because it starts ZuulWeb in a thread, and ZuulWeb becomes unhappy that it can't find its static files - rather than anything actually impacting the test itself 22:23:11 jeblair: cool 22:23:45 I'll probably do that as a followon patch for now so that we don't need another rebase patch flood 22:24:24 mordred: ++. anything else we should know? 22:24:50 don't think so - mostly having people grok the docs and play with the build/run commands to make sure things make sense 22:25:22 I've also started on a similar patch for our friends with storyboard-webclient - since the tooling in use there is actually deprecated 22:25:48 * jhesketh adds topic:zuul-javascript-tooling to his review queue 22:25:58 should we expect all this tooling to be deprecated in 6 months too :) 22:26:00 ya -- if folks can play with that this week, it's probably a good idea to get lots of devs trying it out since we don't all use mordred's laptop 22:26:01 but nobody here needs to necessarily review that - just saying trying to apply the learning from here to over there so that those of us who look at both can look at a thing that's the same 22:26:06 looks awesome mordred (& whoever else worked on it) 22:26:29 jhesketh: thanks! remind me to tell you some stories about all the things I've learned about javascript frameworks, btw 22:26:42 jhesketh: i think feverish-mordred also worked on it, but hopefully all his code has been replaced :) 22:26:43 sounds like a great beer conversation 22:26:43 jeblair: oh - one more thing ... 22:27:33 there are a couple of diffrent approaches we can take to some of how this works, I've picked one that made the most sense to me at the time - but I think that once people poke at this a bit having a discussion to make sure we're happy about that would be nice to have with the team 22:28:32 most notably as it relates to the intersection of html and javascript - webpack happily supports having no html files, generating small stubs and then building the html in javascript directly, which is similar to how status works 22:28:34 * SpamapS has to flip to a diff meeting now 22:28:34 mordred: are there notes about other options in the changes themselves, or should we just have a general discussion after the holidays? 22:28:52 SpamapS: o/ 22:28:55 then angular has a notion of html templates that are being used in builds and jobs which require us to have at least some amont of html in the tree 22:29:16 jeblair: I think we should just have a general discussion - it won't be hugely invasive to change our approach, even in a few months 22:29:43 it's more of a "which worldview do we like, and which one would we, as a team, prefer to focus on" 22:29:54 mordred: sounds good. so as long as this approach works, we can try it out and gain context for that discussion later... 22:29:55 but I think people likely need to get their hands dirty first 22:30:02 jeblair: ++ 22:30:25 mordred: fwiw I found the angular to be really easy to understand as a non front end webby person 22:30:41 clarkb: ++ i like this approach to angular 22:31:00 clarkb: me too - although I now finally understand what's going on in the other angular too - but I don't think I would hav egotten there without tristanC's excellent work 22:31:58 mordred: thanks for all this work, i'm excited :) 22:32:13 jlk: you want to update us on your latest adventures with github3.py? 22:32:32 sure! 22:33:04 So github3.py is going through a refactor: https://github.com/sigmavirus24/github3.py/issues/670 22:33:23 and the upstream author would like that to conclude before making a release 22:33:40 there's a 'work items' list in that issue, btw 22:34:04 have to scroll down to it, i don't see a way to deep-link 22:34:10 I've offered to help with that work, and upstream took me up on it. I now have merge / review rights, and my presence has spurred upstream to start work again 22:34:38 of note, https://github.com/sigmavirus24/github3.py/pull/747 landed just recently, my first stab at a work item 22:35:00 now I can take on further items from that work list, and together we may get this done relatively soon. (weeks, given the holiday). 22:35:06 sweet! 22:35:48 I also ran across another library https://github.com/PyGithub/PyGithub which may be worth looking into, if github3.py work draws out. 22:36:07 and I dug into what it would take to just do our own thing with requests. 22:36:38 It'd be a fair amount of work to do our own thing, re-implementing some things like iterating over paginated returns and such 22:37:19 yah - those are where all the fun is 22:38:00 "easy" enough to copy/paste, but that's kinda shitty behaviour 22:38:35 anyway, that's where things are. I'm going to keep banging on github3.py this week, and some of next week around the holiday. 22:39:25 jlk: do you have a good sense for whether that work list is a month, several months or a year? 22:39:38 (now that you're diving in and landing stuff) 22:40:12 Is there some reason a release can't happen in a sort of stable-bugfix fashion? 22:40:46 mordred: Im' guessing weeks, not months 22:41:38 SpamapS: well, it'd be "no features, except for this one we really want" type release. Might be possible, but I wanted to try this path first 22:41:39 Like the refactor is 0.9.6, but I see there's a 1.0.0a4 .. can we maybe help get a 0.9.7 out with the fixes we need? 22:42:24 jlk: cool. so it sounds like you jumping in and helping should potentially let us cut a release without taking on the cost of any work-around behaviors 22:42:33 that's my hope 22:43:23 SpamapS: I also hoped to gain some good will, and show some good faith, before asking for special treatment, or pushing to be allowed to do a point release (with our needed feature). 22:44:14 Yeah sounds like Ian's stretched a bit thin on it. 22:44:22 jlk: ++ that seems like the nice thing to do. 22:44:35 sounds like maybe we stay on this course for now, and check in after holidays about whether we should continue, beg for a release, or change to pygithub. 22:45:23 sounds fair 22:45:31 also, i'm assuming if anyone wants to pick up one of those work items and earn more brownie points, they are available :) 22:45:37 ++ 22:46:32 I can help anybody who wishes to do that. 22:46:45 jlk: thanks for diving in! yay cross-project collaboration! 22:47:05 \o/ 22:47:22 i want to update folks on the zuul-ci.org work i've been doing 22:47:39 the openstack-infra spec to host such sites was approved 22:47:52 and i'm very close to having both dns and mailing list hosting implemented 22:48:21 * mordred hands jeblair a pie 22:48:34 i'm optimistic that shortly after the holidays, at the latest, we'll be able to have zuul-specific mailing lists at lists.zuul-ci.org 22:49:31 jeblair: have you had a chance to go through my reviews on the mailman stuff? 22:49:48 clarkb: not yet 22:50:22 clarkb: are they 'fix small things' or 'it will never work' reviews? :) 22:50:31 fix small things 22:50:47 cool; i'll switch back to that as soon as i get the dns stuff up 22:50:50 and I think we need to semi schedule a cut over as the config has laready drifted enough that some things were missed in the update openstacl lists list 22:51:07 folks may have noticed an openstack-infra/zone-zuul-ci.org repo 22:51:16 that will contain the dns entries for zuul-ci.org 22:51:35 i'm expecting that to be managed by us -- probably just give zuul-core rights on that? 22:51:52 (though, i'm also not expecting it to change very often) 22:52:29 * SpamapS returns from other lame meeting 22:52:36 or, zuul-core + infra-core to be more specific 22:52:49 very excited to hear there's a zuul website going up! 22:53:05 well no A records in the zone when I last read it :) 22:53:14 but this is step 0 to that 22:53:16 clarkb: who needs A records :) 22:53:31 yeah, that's like step 3 or 4. it is in the plan. :) 22:54:20 * SpamapS wants AAAA records 22:54:30 SpamapS: they will be there :) 22:54:44 SpamapS: along with RRSIG :) 22:54:52 pabelanger: have you had a moment to work on the github reporting? 22:55:46 also, in parallel, since we're almost out of time -- i think the next time this meeting is scheduled not on a holiday is january 8, 2018 22:56:13 shall we plan on skipping until then? 22:56:32 I will not be around on a monday until then 22:56:54 the 2nd we will have the infra meeting though so can talk important things there if they come up 22:57:53 clarkb: ++ like how much our heads still hurt 22:58:21 #info next meeting january 8, 2018 22:58:21 exactly 23:00:42 thanks everyone! 23:00:56 have happy holidays! :) 23:01:01 #endmeeting