16:00:01 #startmeeting api-sig 16:00:01 Meeting started Thu Sep 6 16:00:01 2018 UTC and is due to finish in 60 minutes. The chair is cdent. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:02 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:04 The meeting name has been set to 'api_sig' 16:00:05 #link agenda https://wiki.openstack.org/wiki/Meetings/API-SIG#Agenda 16:00:15 #chair edleafe elmiko dtantsur 16:00:16 Current chairs: cdent dtantsur edleafe elmiko 16:00:20 who is here today? 16:00:35 o/ 16:01:15 o/ 16:01:23 not me 16:01:25 just finishing another meeting 16:02:07 #topic oldbiz 16:02:26 #link minutes http://eavesdrop.openstack.org/meetings/api_sig/2018/api_sig.2018-08-30-16.00.html 16:03:06 edleafe was to keep an eye on this api 3.0 stuff 16:03:07 he did 16:03:08 huzzah 16:03:15 yay me! 16:03:21 cdent reorder some of his pending guidelines which somehow have bad dependencies 16:03:23 i did 16:03:45 yeah, i was really curious about the openapi discussion that went on last time, but never got a chance to read the minutes. is there a quick summary? 16:04:37 elmiko: just that it doesn't seem like anything we need to act on now 16:04:47 ack, thanks 16:04:50 But to keep an eye on it in case it develops momentum 16:05:36 i have a selfish personal interest in the openapi stuff 16:05:52 there was also an unnamed action to "reinvigorate version discovery and consumption guidelines", but I don't think that went anywhere 16:06:05 i've been working on a tool for arbitrary code generation from openapi, no support for 3.0 yet, but i'm hoping to add it soon 16:06:21 * dtantsur does not quite believe in code generation from schemas 16:06:31 elmiko: interesting. How would such a tool be useful in the OpenStack community? 16:06:32 that's a good way to put it dtantsur 16:06:55 edleafe: i'm not sure that it would unless folks wanted to use more openapi for defining their api's 16:07:00 dtantsur: he did say "arbitrary", not "useful" :) 16:07:23 lol, okay 16:07:30 but i've tried to build something that has a wide margin of applicapability by allowing the user to generate templates that get applied against the openapi data 16:07:51 in essence you have the freedom to customize how the code is generated, which is something that current tools kinda lack 16:08:06 dtantsur: what's not to believe in, it works 16:08:20 elmiko: sure, I did not use "believe" in that sense 16:08:26 hahaha =) 16:08:40 now is it /useful/? that's a different question 16:08:46 based on my rust-openstacksdk experience, a literal SDK generated from our API would be quite unusable 16:08:50 anyways, if folks are curious, https://gitlab.com/elmiko/deswag 16:09:07 my favourite example includes using empty string for missing fields (hello nova) 16:09:15 * examples include 16:09:17 dtantsur: using a tool that makes all the code choices for you, i tend to agree 16:09:59 I'd be happy if something could generate https://github.com/dtantsur/rust-openstack/blob/master/src/network/protocol.rs#L198-L236 16:10:22 but even this low-level structure required some conscious choices 16:10:36 elmiko: I'm not trying to diminish the value of your work, just sharing my experience :) 16:11:22 dtantsur: no worries, i think the field of auto-generated code has made some really bad steps in the past and that has left many people (myself included) highly skeptical about their value 16:12:03 edleafe: to your question earlier, i'm not sure that openapi and code generators can necessarily help projects in openstack that already have established code bases 16:12:13 i think it's more attractive for projects that are starting up 16:12:17 when ed and I are at the ptg we'll be sure to make code generation plans that ignore both elmiko and dtantsur 16:12:27 :D 16:12:27 ++ 16:12:36 elmiko: no last-minute travel approval? 16:12:43 heh, sadly no 16:12:46 not _yet_ 16:12:49 LOL 16:13:07 elmiko: I'm still thinking of generating some rust, then fixing it manually 16:13:12 but it requires writing a generator :) 16:13:16 my last-minute travel plans are trying to figre out which train i will take to come visit cdent =D 16:13:23 \o/ 16:13:46 dtantsur: check out my project, you can make jinja templates that create code for you from the openapi schema. it might give you the flexibility you need 16:14:08 I certainly will, await pull requests :) 16:14:13 also, it's very alpha currently. so yeah... 16:14:17 hehe ++ 16:15:03 I wish we had an example schema for at least one project 16:15:07 (one microversion of one project) 16:15:23 this conversation could be much more constructive if we had something real 16:15:27 I guess gilles was gonna do that, but then got distracted by graphql 16:16:24 we might be able to transpile one of the sphinx schema things into openapi 16:17:15 i had thought that's what graham was building up to 16:17:31 graham had a change of circumstances 16:18:04 ah, ok 16:18:14 hope everything is well for him 16:18:54 yes, I think so, just less time for "fun" openstack things 16:19:02 ack 16:19:24 Shall me move on or is there more to say about openapi and graphql? 16:19:35 nothing more from me 16:19:50 put a fork in it 16:20:00 #topic open mic and ongoing or new biz 16:20:01 #link https://etherpad.openstack.org/p/api-sig-stein-ptg 16:20:13 so hey, monday we gonna meet up with some folk and chat about some stuff 16:20:28 that etherpad is sort of an agenda, but more than likely we'll talk about whatever people in the room want to talk about 16:20:33 do we need to drive it more than that? 16:21:01 i think what is proposed sounds good 16:21:48 hopefully you'll have a nice crowd and the agenda will set itself ;) 16:22:05 We have microversions mentioned, so we should have a good crowd 16:22:13 hahaha 16:22:21 Add something about k8s, and it'll be standing-room only! 16:22:24 you'll have a crowd, not sure how "good" it will be XD 16:22:30 ooh yeah, good point 16:22:43 and edge, edge! 16:22:47 serverless! 16:22:53 blockchain! 16:23:19 i think we might all be getting a bit jaded 16:23:23 "Running your serverless microversioned blockchain in Kubernetes" 16:23:25 hahaha 16:23:32 cdent: you /think/... 16:23:35 XD 16:23:51 * cdent calls the VCs about edleafe's idea 16:24:00 ++ 16:24:24 I thought that this was under NDA 16:25:16 * cdent points and laughs 16:25:37 * edleafe feels like the Winklevoss twins 16:25:56 haha 16:27:00 :D 16:27:07 Anything else on the ptg? 16:27:30 have fun and represent! 16:27:35 We'll save a seat for elmiko 16:27:45 'Cause you know he'll show up 16:27:49 =D 16:27:53 good planning 16:28:04 * elmiko would love to come strolling in on monday 16:29:12 elmiko is playing the long con 16:29:29 #topic guidelines 16:29:29 #link https://review.openstack.org/#/q/status:open+project:openstack/api-sig,n,z 16:29:30 * edleafe isn't fooled 16:29:50 I don't think we've had any changes, other than me unblocking some stuff that got stacked when it didn't need to 16:30:33 obpromise: more work on guidelines soon! 16:31:05 \o/ 16:31:11 #topic bug review 16:31:11 #link https://storyboard.openstack.org/#!/project/1039 16:31:24 no changes here either 16:31:32 what is it this sig does, anyway? 16:32:24 chats with awesome people once a week? 16:32:27 we hold some of the best meetings in the entire openstack community damnit! 16:32:28 make jokes 16:32:35 see, solidarity! 16:32:39 =D 16:32:44 word 16:32:59 community building 16:33:24 #topic weekly newsletter 16:33:25 #link https://etherpad.openstack.org/p/api-sig-newsletter 16:33:39 whoever does the letter (not me), can you say there will be no meeting next week? 16:33:57 Sure, I'll do it 16:34:04 thanks edleafe ! 16:34:34 huzzah 16:34:44 I'll send the usual ping in -sdks 16:34:52 huzzah 16:34:58 anything else from any one? 16:35:17 pour one out for me at the ptg, i'll miss you guys =) 16:35:23 yeah yeah 16:35:26 hahaha 16:35:33 * cdent remembers take extra cash for elmiko's drinks 16:35:35 * edleafe thought you were gonna say "roll one up..." 16:35:41 edleafe: ++ 16:35:49 It *is* Colorado, after all 16:35:54 tru dat 16:36:04 and a beer or two for me 16:36:16 ++ 16:36:19 headline: api-sig lights up masive doobie, decides REST sucks, goes back to xmlrpc 16:36:26 LOL 16:36:37 and with that 16:36:39 #endmeeting