15:59:47 <etoews> #startmeeting api wg
15:59:48 <openstack> Meeting started Thu Jun  9 15:59:47 2016 UTC and is due to finish in 60 minutes.  The chair is etoews. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:59:50 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:59:53 <openstack> The meeting name has been set to 'api_wg'
15:59:55 <elmiko> hi
15:59:58 <etoews> hi
16:00:45 <cdent> hola
16:00:47 <rosmaita> hello
16:01:05 <tsymanczyk> hiya
16:01:19 <etoews> #link https://wiki.openstack.org/wiki/Meetings/API-WG#Agenda
16:01:44 <etoews> #topic previous meeting action items
16:01:52 <etoews> #link http://eavesdrop.openstack.org/meetings/api_wg/2016/api_wg.2016-06-02-16.00.html
16:01:59 <etoews> nada
16:02:23 <etoews> actually doing work during the meetings is awesome
16:02:28 <cdent> did it all during, yeah
16:03:17 <etoews> let's move the weekly newsletter until later. we have some guests so let's find out what they're up to.
16:03:37 <etoews> #topic Glance v2 image sharing enhancement
16:03:39 <elmiko> sorry, i'm kinda bifurcated here. dealing with a local fire
16:03:51 <etoews> #link https://etherpad.openstack.org/p/newton-image-visibility-changes-api
16:04:01 <etoews> rosmaita: that's you right?
16:04:13 <rosmaita> yep
16:04:31 <rosmaita> please take a quick look at the etherpad
16:04:46 <rosmaita> otherwise i will be typing exactly what's there in here
16:05:00 * cdent looks
16:05:07 <rosmaita> but let me kno wif you need more cotext
16:05:12 <rosmaita> *context
16:09:37 <cdent> rosmaita: the description, in just the context of itself, makes sense
16:10:04 <rosmaita> ok
16:10:29 <rosmaita> also, i should introduce tsymanczyk who is here, he is the author of the spec
16:10:55 <tsymanczyk> o/ hi again
16:11:11 <cdent> I'd go for the new workflow, but I'm generally not a good person for questions that involve issues of backwards compatibility because a) I'm not a hugely active user/deploy of glance (or other openstack apis), b) I'm a cruel person who thinks it is okay that upgrades change things, c) Glance has a history of people being easily upset about API changes :(
16:11:50 <rosmaita> yes, (c) is a real problem
16:12:24 <rosmaita> it's only a slight backward incompatability
16:12:42 <rosmaita> what do you think etoews , elmiko  ?
16:12:57 <etoews> i'm leaning towards new workflow.
16:13:07 * elmiko looking at etherpad
16:13:19 <etoews> let me ask you this, do you know exactly who the users are who rely on old sharing workflow?
16:13:28 <etoews> to-be-old...
16:13:51 <etoews> what percentage of total users do they make up?
16:14:04 <rosmaita> hard to say
16:14:16 <rosmaita> one user is the rackspace cloud control panel
16:15:12 <cdent> rosmaita: is there much PATCH deployed in Glance already?
16:15:15 * rosmaita is thinking how to break the news to them (i work in the same office)
16:15:19 <elmiko> the new workflow seems like an improvement, imo, but yeah the backward compat issue. i'm not sure about that
16:15:34 <rosmaita> cdent: yes, it is the only way to update an image record in v2
16:15:41 <nikhil> there may be some in other public clouds but don't expect too many
16:16:42 * nikhil apologizes for making a theatrical appearance in the meeting
16:17:00 <rosmaita> cdent: in v2, you change visibility using PATCH now (it's just that there are only 2 values)
16:18:23 <cdent> rosmaita: is that well received. I don't mean PATCH itself, but the use of JSONPatch. This isn't really directly related, I'm just curious if that itself has gone well, because some people are not keen.
16:18:59 <rosmaita> i guess that's a good question for nikhil, i haven't heard any complaints about it
16:19:09 <nikhil> yes and no
16:19:15 <cdent> nikhil: It's only theatrical if it has a musical cue
16:19:17 <etoews> a lot of the sdk people i know are not fond of it
16:19:34 <nikhil> dan-dan-dan
16:19:35 <nikhil> lol
16:19:38 <nikhil> people are usually quite about it unless someone wants to write a client
16:19:58 <nikhil> yeah, what etoews said (I tried to generalize it)
16:20:09 <nikhil> sdk, osc folks are usually not happy
16:20:34 <nikhil> we also had something called as a warlock lib that does schema validation-- this is very painful for clients
16:21:00 <nikhil> s/had/have/
16:21:28 * nikhil stops giving feedback that's not asked for
16:22:24 <cdent> rosmaita: the semantics of the described stuff sound good, the social (?) issues are harder to contemplate and comment upon...
16:22:36 <cdent> where "sound good": consistent with reality
16:23:09 <etoews> does glance dance the microversion?
16:23:17 <rosmaita> no
16:23:19 <nikhil> not yet
16:23:25 <nikhil> it'd have been so much easier
16:23:29 <etoews> ya...
16:23:48 * nikhil plans to find a weekend for that work :)
16:24:04 <etoews> but i'm definitely *not* suggesting "do microversions before doing this"
16:24:06 <cdent> if microversions make versioning easier, does that mean that nanoversions make it ever easier?
16:24:15 <cdent> s/ever/even/
16:24:15 <rosmaita> lol
16:24:19 <nikhil> lol
16:24:25 <etoews> naturally :)
16:24:34 <nikhil> cdent: I for some reason have been feeling the same thing
16:24:34 <rosmaita> with picoversions we wouldn't have to do anything at all
16:24:45 <nikhil> femto?
16:25:03 <rosmaita> i thnkn femtoversions write themselves
16:25:12 <cdent> rosmaita++
16:25:16 * nikhil rofl
16:25:35 <etoews> our job here is done
16:26:01 <elmiko> wow... femtoversions lol
16:26:47 <rosmaita> ok, so to summarize : going to the new workflow isn't opposed by the API WG,but they point out that some glance consumers may be outraged, and glance will have to deal with the fallout
16:27:01 <cdent> that's a reasonble summary
16:27:03 <etoews> +1
16:27:36 <rosmaita> ok, thank you
16:27:43 <etoews> i think glance operators can mitigate the outrage in some cases so that's a big plus
16:27:43 <nikhil> rosmaita: I think we can reduce that pain and get some scoped feedback separately from those stakeholders.
16:27:57 <rosmaita> nikhil: +1
16:28:23 <etoews> it might be worthwhile to mention a strategy for outrage mitigation in the spec
16:28:31 <rosmaita> :)
16:29:03 <cdent> all specs probably should have that, at least in the template
16:29:12 <rosmaita> tsymanczyk: ^^
16:29:22 <tsymanczyk> enraged? 1 bourbon 1 scotch 1 beer.
16:29:40 <rosmaita> on that note, i also have the next agenda item
16:29:45 <tsymanczyk> still enraged? goto step 1
16:29:51 <rosmaita> :)
16:30:14 <nikhil> well said :)
16:30:29 <cdent> rosmaita: glare?
16:30:34 <rosmaita> yep
16:30:35 <nikhil> lol
16:30:44 <nikhil> oh wow, interesting
16:30:46 <etoews> the image glare
16:30:55 <rosmaita> mfedosin asked me to bring the latest patch of the glare API to your attention
16:31:01 <etoews> #topic updated Glare API spec available for review
16:31:07 <etoews> #link https://review.openstack.org/#/c/283136/
16:31:10 <cdent> I remember we looked at this back in february
16:31:31 * nikhil remembers
16:31:45 <rosmaita> i'm not sure how much has changed, mostly clarifications in the spec, i think
16:32:14 <rosmaita> but we'd appreciate a lookover ov the latest version if you have time
16:32:27 <rosmaita> that's all from me
16:32:28 <tsymanczyk> need to go commute. thanks for the feedback everyone.
16:34:53 <nikhil> the api section is here for reference:
16:34:57 <nikhil> #link http://docs-draft.openstack.org/36/283136/6/check/gate-glance-specs-docs/ec6fccb//doc/build/html/specs/newton/glare-api.html#rest-api-impact
16:35:03 <etoews> if glare was microversioned, would it still need the /v1 in the URL?
16:35:07 <cdent> can an artifact change type while still effectivey being the same thing?
16:35:09 <etoews> i think not
16:36:14 <etoews> my "i think not" applies to my own comment about microversion
16:36:53 <nikhil> cdent: can you please elaborate?
16:38:03 <cdent> nikhil: not really, it was sort of a question to toss out to see if I could get a better idea of what "artifact type" means. Back when we first looked at this there was some confusion (especially from me).
16:38:56 <nikhil> cdent: artifacts being a generic way to store data assets -- some that may vary in their definition like heat templates and murano packages
16:39:12 <nikhil> cdent: so, the type is 'heat_template' or 'murano-package'
16:39:24 * cdent nods
16:39:41 <nikhil> cdent: defined in the source tree and having such limited capacity of fields on them
16:39:52 <cdent> can their only be one blob per artifact?
16:40:06 <nikhil> cdent: so for heat templates, there will be property H1, H2, H3 and for murano package it will be M1, M2, M3
16:40:08 <cdent> (I'm skimming the spec instead of reading, I suspect some of this is in there)
16:40:22 <nikhil> yeah
16:40:35 <cdent> okay, thanks
16:40:59 <nikhil> I don't have exact answer on what blobs are: but generic answer is yes
16:41:15 <nikhil> it's a part of the artifact definitions
16:42:24 <cdent> yeah, I'd like to roll back to what etoews asked: is /v1 a requirement if microversions are going to be used? And: If there is any plan to ever use microversions, then they need to be built in from the start. (I say this as someone who is not really a fan of microversions, but see their utility).
16:42:51 <etoews> i was just about to bring that up again myself ;) thx
16:43:15 <etoews> imo, glare should microversion from the start and drop the /v1
16:43:58 <nikhil> Thanks for that feedback, really useful to know.
16:44:20 <rosmaita> we will pass that along to mfedosin and see what he thinks (about the microversions)
16:44:39 <elmiko> +1 on what cdent and etoews have said
16:44:55 <rosmaita> though it's probably worth putting that comment on the spec for posterity
16:45:06 <etoews> rosmaita: yep. i'll comment right now.
16:45:12 <rosmaita> cool, ty
16:45:35 <etoews> anything else on this topic?
16:45:42 <rosmaita> not from me
16:45:44 <nikhil> none from me
16:45:58 <cdent> I'll be given another review later when I have a bit more ability to read with focus
16:46:04 <cdent> giving
16:46:05 <cdent> sigh
16:46:13 <etoews> cdent: elmiko: are we skipping the newsletter this week or hammer it out now?
16:46:20 * cdent needs femtoversions to fix his brain
16:46:32 <elmiko> well, i think as a rule we should not skip it
16:46:54 <elmiko> but, i say that as someone who's already maxed on bandwidth... so, grain of salt
16:46:59 <cdent> etoews: we can probably make one now, I don't think anything has changed so we can push it out and not skip
16:47:08 <etoews> okay.
16:47:08 * nikhil thanks API WG for the feedback, leaves meeting.
16:47:18 <etoews> thanks nikhil rosmaita !
16:47:21 <cdent> elmiko: I'll launch today's
16:47:25 <etoews> #topic send the weekly newsletter
16:47:25 <etoews> #link https://wiki.openstack.org/wiki/API_Working_Group_weekly_email_template
16:47:25 <etoews> #link https://etherpad.openstack.org/p/api-wg-newsletter
16:47:26 * rosmaita waves bye
16:47:27 <elmiko> thanks cdent !
16:47:34 <etoews> thanks cdent !
16:47:49 <etoews> if i don't comment on the glare api now, it'll never happen.
16:48:08 <elmiko> the new stuff i'm working on is eating more and more of time, and sadly openstack is getting the short end of the stick =(
16:49:25 <cdent> a common (and sad) story
16:49:44 <elmiko> yeah...
16:49:51 <elmiko> i just want to be honest about it
16:51:50 <etoews> elmiko: have you found anyone that might be able to take your place?
16:52:06 <elmiko> sadly no, the folks i asked were all also too busy lol
16:54:55 <cdent> writing in the newsletter makes me think: do need a policy on when/how to abandon guidelines whose authors are not reacting to feedback?
16:55:45 <elmiko> hmm, not a bad idea. but it does add more process
16:56:11 <cdent> I've made some small edits (mostly just putting in a different, chatty, intro paragraph.
16:56:13 <rosmaita> i am guilty of that
16:56:30 <cdent> If you both can give it a once over, I'll make it go
16:56:32 <rosmaita> have not worked on the "versions" guideline in like 6 months
16:57:59 <elmiko> lgtm, i like the note about the glance conversation. keeps it very topical
16:58:51 <cdent> etoews: review on https://etherpad.openstack.org/p/api-wg-newsletter ?
16:59:32 <etoews> ship it
16:59:41 <etoews> #endmeeting