16:02:53 #startmeeting api-sig 16:02:53 Meeting started Thu Feb 1 16:02:53 2018 UTC and is due to finish in 60 minutes. The chair is cdent. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:02:54 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:02:56 The meeting name has been set to 'api_sig' 16:03:03 #chair dtantsur elmiko edleafe 16:03:04 o/ 16:03:05 Warning: Nick not in channel: elmiko 16:03:06 Current chairs: cdent dtantsur edleafe elmiko 16:03:18 sorry about the slight delay, lost track of time 16:03:33 #link Agenda https://wiki.openstack.org/wiki/Meetings/API-SIG#Agenda 16:03:52 #topic old biz 16:03:56 there was no meeting last week 16:04:04 and the week before had no action items 16:04:17 #topic new biz 16:04:38 #link action apis thread: http://lists.openstack.org/pipermail/openstack-dev/2018-January/126334.html 16:05:14 a chance to say what we think about such actions? :) 16:05:16 that ^ was a thread on the dev list that invited our input on dealing with "actions" in uris. our input was invited but I think everyone was sufficiently distracted that they mimssed it 16:05:28 sure, if you like 16:05:59 probably not overly constructive though 16:06:42 perhaps, but I do think the thread deserves a bit of attention, even though it is somewhat old now 16:06:44 \o (sort of) 16:06:56 o/ 16:07:04 sorry, missed the calendar bing 16:07:06 * edleafe is on a phone meeting 16:07:21 elmiko: no worries, the only reason I remembered is because dtantsur was on the ball 16:07:39 dtantsur++ 16:07:52 * elmiko pulls up agenda 16:07:56 we were discussing how this thread http://lists.openstack.org/pipermail/openstack-dev/2018-January/126334.html invited our attention 16:08:03 be we haven't given it any (yet) 16:08:19 I kinda agree that if we do actions, we should make them explicit in the URL 16:08:23 ooh, action resources, my favorite =) 16:08:46 because, discoverability, all the things 16:09:35 dtantsur: like explicit url to a resource based action? 16:09:57 elmiko: like, using //actions/ not just //actions 16:10:04 ack 16:10:08 I mean, both are evil, but the former is a lesser one IMO 16:10:12 yeah 16:10:14 so are they saying "let's not get tied up with this RESTy stuff"? 16:10:48 i.e., don't use POST /resource with an action in the body? 16:10:48 edleafe: not sure I get you, but they're picking between several non-REST approaches :) 16:10:58 imo, i like the idea of tightly scoped urls to define the action, but then having the actions be resources based for tracking and asynch behaviors 16:11:16 dtantsur: yeah, that's my point. They are considering everythine *except* the REST way 16:11:21 right 16:11:47 which gets us down the rabbit hole of how to express e.g. a server stop in a restful way :) 16:12:04 we track these sort of things as tasks, which are resources in their own right 16:12:12 mugsie++ 16:12:14 mugsie: right, this is the best approach IMO 16:12:26 https://developer.openstack.org/api-ref/dns/#zone-import -> https://developer.openstack.org/api-ref/dns/#zone-ownership-transfers-accepts 16:12:28 but if we're speaking about existing APIs, this means a complete rewrite 16:12:57 or ... 2 ways of doing the same thing at least temporarily 16:13:11 i don't think we should accept the non-resource based style for a guideline simply because it's a lot of old code 16:13:30 mugsie: what is the conceptual difference between 'action' and 'task'? 16:13:40 i know it's a tough pill to swallow, but we can't advise folks to use the non-resource methodology 16:14:01 edleafe: nothing - it was just our word choise 16:14:08 choice* 16:14:16 elmiko: I agree. I'm trying to imagine how we can lead this conversation to not end up with "- You should do X. - No way" 16:14:30 mugsie: ok, that's what I thought 16:14:46 but we diud make sure we kept track of old tasks, and they all get an ID, and have a status that progresses as the task /action is performed 16:15:05 dtantsur: ack 16:15:27 that's a tough problem to solve, i'm not sure how we make a good guideline without upsetting some folks 16:15:29 elmiko: I think the approach should be: "Use this RESTful approach. Until you can get there, this non-RESTful approach is the least sucky, so try for that" 16:15:37 edleafe++ 16:15:44 yes, give folks a way to get there 16:15:58 Does someone want to volunteer to join the thread? 16:16:05 IOW, never endorse a non-RESTful approach as our recommendation 16:16:13 or even some ones 16:16:15 edleafe++ 16:16:22 edleafe: that's what I would say, yes 16:16:30 cdent: I guess I can reply 16:16:31 cdent: i can join, might not be today though 16:16:35 I can jump on it too 16:16:39 \o/ 16:16:44 well, let 16:16:46 ugh 16:16:52 let's not repeat ourselves 16:16:55 ++ 16:17:01 we should really just have one API-SIG response 16:17:10 that's fair 16:17:25 #action edleafe to jump on the email thread 16:17:50 Any other new biz or open discussion? 16:17:56 cdent: with my knees, I think I'll step lightly :) 16:18:13 it's a soft landing 16:18:25 * elmiko chuckles 16:18:27 One open discussion topic might be "oh hey, that PTG thing is coming round again" 16:18:37 \o/ 16:18:46 * dtantsur has applied for a visa.. fingers crossed 16:19:07 you need a visa dtantsur ? bummer 16:19:14 anybody definitely not going? 16:19:29 cdent: sucks to have such a passport :( 16:19:45 cdent: i don't think i'll be there 16:19:48 * dtantsur mumbles something about the foundation considering visas when deciding on a venue 16:19:56 I'm coming, but IBM is dragging its heels on approving 16:19:59 elmiko: nooooooooooo! 16:20:01 although, i'm willing to video call in if folks would like 16:20:04 Already bought airline tix 16:20:15 sorry cdent =( 16:20:21 elmiko: it may work depending on the room equipment 16:20:27 elmiko-- 16:20:28 elmiko: video Guinness isn't as good 16:20:30 cool 16:20:35 haha, yeah 16:20:38 heh 16:21:07 do we know what days are the api-sig stuff? 16:21:15 yeah, one sec 16:21:16 (maybe i can make a case for a short journey) 16:21:45 #link ptg sched: https://docs.google.com/spreadsheets/d/e/2PACX-1vRmqAAQZA1rIzlNJpVp-X60-z6jMn_95BKWtf0csGT9LkDharY-mppI25KjiuRasmK413MxXcoSU7ki/pubhtml?gid=1374855307 16:21:49 Monday? 16:21:59 yeah, only a day this time 16:22:06 elmiko: isn't the airline tix the biggest cost? Or are they worried that you will be slacking off on your day job? 16:22:12 fine, I'll have time to check out the city 16:22:34 on Tue, I mean 16:22:46 without microversions to discuss, we just don't have that much to say... 16:22:52 edleafe: little of column a, little of column b 16:23:07 cdent: what do you mean by "without"? ;) 16:23:14 we ALWAYS have microversions to discuss 16:23:14 and not so much slacking, just that our group doesn't do any openstack anymore 16:23:20 haha 16:23:28 elmiko: so the $$ will be the same. You can just work remote on Tues, Wed, etc 16:23:33 * dtantsur gently reminds about his SDK guideline to bike shed on 16:23:54 edleafe++ 16:24:00 i'll take one more pass at getting approval 16:24:05 dtantsur: I was meaning more that the great existential battles about microversions are behind us 16:24:15 * dtantsur is less optimistic 16:24:42 cdent: We could bikeshed on what to rename "microversions" to! 16:24:44 if we ever try to push them to more projects, the past battle will seem just a shadow of the horrible present 16:24:53 the battles weren't won, people just got fatigued and decided there were better things to do 16:24:53 edleafe: !!! 16:25:56 do I remember it right that somebody proposed "microversion everything" as an openstack-wide goal? 16:26:51 I think it has been on the etherpad, but never made it past that 16:26:55 dtantsur: I think that was suggested, but goals are supposed to be achievable in one cycle 16:27:03 .. or ever ;) 16:27:19 okay then. if it ever becomes serious, we're gonna have a new round. 16:27:21 and I suspect some projects would just outright refuse to do it :P 16:27:49 It does seem odd that the way to ensure that you don't break an API (via microversions) is only possible by breaking your API (by switching to microversions) 16:28:05 edleafe: why breaking? 16:28:20 I recall us switching quite smoothly (well... from this standpoint) 16:28:42 dtantsur: most projects that have objected have raised breakage as a reason for not wanting to implement 16:28:45 we just declared whatever we had 1.1 or something like that 16:28:55 curious 16:29:08 I was objecting because I find it confusing and error-prone 16:29:09 Some have v1, v2, etc 16:29:41 we had v1, so we just froze it at the point of switch. then we can v1+microversion for newer things 16:29:47 but this is side-tracking a bit, sorry :) 16:30:16 i wonder if the backward compatiblity nature of microversions scares ppl as well? 16:30:16 regarding the ptg, we should probably prepare some topics 16:30:23 ++ 16:30:28 how about actions? 16:30:34 actions++ 16:30:39 let's do that at 16:30:43 #link ptg etherpad https://etherpad.openstack.org/p/api-sig-ptg-rocky 16:30:47 and I'm quite serious about microversions in SDKs, it's unlikely we merge them 16:31:11 and I'd probably meet people from SDKs before we do anyway 16:33:30 I think with SDKs we're going to have the same problem as with "capabilities" 16:33:47 There are multiple levels that are all being called the same thing 16:34:02 dtantsur tried to separate the definitions 16:34:30 but it does seem that there is more of a spectrum of SDK focus 16:35:07 It's question that's come up a lot, so we can do a service by providing some help with making it a real conversations 16:35:18 #topic guidelines 16:35:32 #link pending guidelines https://review.openstack.org/#/q/status:open+project:openstack/api-wg,n,z 16:36:07 doesn't look like anything is ready for freeze, conversations continues on dtantsur's thing (a good starting point for the conversation) 16:36:14 the schema thing is languishing a bit? 16:36:18 re: ptg, seems like i'm getting some traction, so keep your fingers crossed 16:36:28 \o/ 16:36:36 i said you guys are getting really pushy 16:36:55 mordred's tome will probably come back to life eventually but not clear when 16:37:03 uhoh 16:37:05 yah 16:37:08 elmiko: we haven't even started ;) 16:37:11 I keep getting distracted by other things 16:37:15 mordred: o/ anything to discuss at the PTG? 16:37:16 * elmiko chuckles 16:37:21 elmiko: I think you can tell your boss you owe me a (very expensive) beer 16:37:30 cdent: LOL 16:37:32 indeed 16:37:40 dtantsur: I'd loe to chat with you about the sdk/microversion stuff at least 16:37:59 mordred: totally! I've proposed a topic on https://etherpad.openstack.org/p/api-sig-ptg-rocky 16:38:06 and would also like to talk with y'all and maybe public cloud folkks about the cloud profile thing I need to update 16:38:09 dtantsur: woot 16:38:09 if they don't send me to hell with my visa request ofc.. 16:38:23 mordred: https://etherpad.openstack.org/p/api-sig-ptg-rocky 16:38:55 * mordred added line 16:39:03 word 16:39:20 #topic bugs 16:39:21 mordred: some details would not hurt 16:39:23 #link bugs https://bugs.launchpad.net/openstack-api-wg/+bugs?orderby=-id&start=0 16:39:34 no new bugs, no progress on existing bugs 16:39:43 stability! 16:39:51 I reckon winter is the hardest api-sig season 16:40:12 yes, especially feature FREEZE 16:40:26 * elmiko facepalm 16:40:30 /o\ 16:40:54 #topic newsletter 16:40:57 #link newsletter https://etherpad.openstack.org/p/api-sig-newsletter 16:41:00 any takers? 16:41:15 I'm doing the email reply, so prolly not me 16:41:21 i can take it 16:41:30 * cdent cookies elmiko 16:42:04 i'll ping folks in like 15-20 16:42:11 anyone have anything to add to the record before I end the meeting? 16:42:53 okay then, thanks everyone for showing up, a pleasure as always 16:42:58 #endmeeting