14:00:41 <ativelkov> #startmeeting glance-artifacts
14:00:42 <mfedosin> o/
14:00:42 <openstack> Meeting started Mon Jun 29 14:00:41 2015 UTC and is due to finish in 60 minutes.  The chair is ativelkov. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:44 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:46 <openstack> The meeting name has been set to 'glance_artifacts'
14:00:53 <ativelkov> Hi folks
14:01:24 <kragniz> hi
14:01:25 <sigmavirus24> Hello everyone
14:01:46 <ativelkov> Nikhil is ooo today, So, I'll run this meeting
14:01:56 <sigmavirus24> Sounds good
14:02:20 <ativelkov> We don't have much agenda for today, as most of the topics are postponed till nikhil's return: https://etherpad.openstack.org/p/glance-artifacts-sub-team-meeting-agenda
14:02:32 <ativelkov> So, just a quick status update on what's going on
14:02:59 * sigmavirus24 listens eagerly as he's been not able to keep up
14:03:09 <ativelkov> we are in the middle of migration from our "declarative framework" to oslo_versionedobjects
14:03:47 <ativelkov> Most important part - the constraints and validation - has to be ported there as they miss that functionality
14:03:58 <ativelkov> I've made that commit last week and it is on review now
14:04:01 <ativelkov> #link https://review.openstack.org/#/c/196041/
14:05:11 <ativelkov> It does not include a complete validation of list/dict fields though, so it has to be added a bit later: for now we almost don't use it in Glance anyway
14:06:44 <ativelkov> The remaining part - a concept of "immutable" field (i.e. the fields which became read-only based on the state of the owner object) probably has to remain at Glance, as it looks very specific to us
14:07:07 <sigmavirus24> That seems like a fair compromise
14:07:48 <mfedosin> what about json schema generation? will it be in glance too?
14:08:27 <ativelkov> As soon as this lands, I'll submit a commit which adds oslo-based way of declaring artifact types in parallel with the old ones. We'll have it reviewed (hope flaper87 get some cycles for that) and then we'll drop the old declarative frmwk
14:09:02 <ativelkov> mfedosin: it depends. As we've decided to get rid of warlock for the client, we (probably) don't need jsonschema at all
14:10:01 <ativelkov> But this may be changed back (if we want to be backwards compatible with v2 we'll have to introduce something similar to v2's property protection,w hich uses schema)
14:10:06 <mfedosin> ativelkov, what will replace it?
14:10:25 <kragniz> when did we decide to get rid of warlock on the client?
14:10:40 <ativelkov> kragniz: last week on this same meeting
14:10:43 <mfedosin> kragniz, on the last meeting
14:10:59 * kragniz must have been asleep
14:11:00 <ativelkov> but that's just a temporary solution for now, just to save some time coding the client
14:11:32 <ativelkov> So, for now we'll have a "static" client which returns objects with attributes corresponding to generic artifact properties
14:12:07 <ativelkov> And all the type-specific attributes placed into "type_specific_proprties" dict of those objects
14:12:13 <mfedosin> you can chack how it's done here https://review.openstack.org/#/c/189860/8/glanceclient/v3/__init__.py
14:12:50 <sigmavirus24> ativelkov: mfedosin is it for everything in the client, or just artifacts in the client?
14:13:11 <mfedosin> just artifacts
14:13:19 <ativelkov> sigmavirus24: well, for now v3 client is going to have only artifacts :)
14:13:46 <mfedosin> yep, tasks and metadefs may use schemas
14:14:25 <ativelkov> This expects the clients of projects which consume particular artifact types (python-muranoclient, python-heatclient etc) to use glance-client to fetch the artifact information and form their proper objects, given that they know the schema
14:14:29 <sigmavirus24> Right
14:14:38 <sigmavirus24> I think kragniz thought warlock was completely gone =P
14:14:54 <ativelkov> I'd prefer it to go, as it is now supported anymore
14:15:09 <ativelkov> Or at least it seems so :)
14:15:27 <sigmavirus24> Yep. Anyway that's a separate topic and I don't want to derail the meeting :D
14:15:52 <ativelkov> Yup, that sounds like a good topic for mid-cycle meetup
14:16:38 <ativelkov> More good news: I participated in last week's IRC meeting of Community App Catalog team
14:16:52 <mfedosin> and they want us :)
14:16:52 <ativelkov> #link http://eavesdrop.openstack.org/meetings/app_catalog/2015/app_catalog.2015-06-25-17.01.log.html
14:17:12 <ativelkov> yup, they are considering Glance V3 to become their backend at some point of time
14:17:43 <sigmavirus24> As you predicted at the summit =P
14:18:04 <mfedosin> sigmavirus24, they had concerns actually
14:18:18 <ativelkov> So, we'll make some kind of PoC for them to demonstrate how Glance handles their current data
14:18:18 <mfedosin> but it's okay now
14:18:37 <sigmavirus24> mfedosin: ah good to know
14:18:49 <ativelkov> We need to align their request with our roadmap - and have them help us to work on API improvements
14:19:13 <ativelkov> Their primary requests are import/export of artifacts and dynamic artifact references
14:19:39 <ativelkov> (i.e. ability to reference artifacts not only by static IDs but also by name and version range)
14:19:48 <mfedosin> nothing of that isn't done yet :(
14:20:13 <ativelkov> mfedosin: well, dynamic references are relatively easy to do
14:20:32 <ativelkov> Unlike import/export
14:20:44 <mfedosin> agree, but import/export is a big task
14:21:14 <mfedosin> since we still have no decision about format
14:21:18 <ativelkov> ivasilevskaya has done some import/export PoC shortly before she left, but it definetely needs much more work
14:22:03 <mfedosin> definitely we need a volunteer :)
14:22:35 * ativelkov voluntols mfedosin
14:22:49 <mfedosin> I'm working on the client currently
14:23:12 <mfedosin> btw, folk,initial commit is ready https://review.openstack.org/#/c/189860/
14:23:20 <ativelkov> Well, we'll see. The client is more important right now, for sure.
14:23:33 <ativelkov> YEs, some reviews for the client code would be appreciated
14:23:33 <mfedosin> please review it when you can
14:23:53 <mfedosin> CRUD part will be done today
14:24:17 <mfedosin> and listing on this week too
14:24:41 <ativelkov> So, the question is, do you guys think we need a CLI for artifacts now, or this can be postponed a bit till we stabilized the API?
14:25:14 <sigmavirus24> ativelkov: I would postpone until we have a stabilized API
14:25:16 <sigmavirus24> That said
14:25:36 <sigmavirus24> Landing the majority of the work and updating it wouldn't be a terrible thing
14:25:46 <mfedosin> sigmavirus24, +1
14:25:50 <sigmavirus24> That means there's a couple large diffs upfront, and then a bunch of small ones
14:25:55 <sigmavirus24> The small ones would be much easier to review
14:25:58 <sigmavirus24> But it's still more work so
14:26:41 <ativelkov> sigmavirus24: thanks, sounds reasonable
14:26:44 <sigmavirus24> If we can keep the reviews under 400 LOC that'd be great.
14:26:57 <ativelkov> Yup, that's a lesson learned from the last cycle
14:27:00 <sigmavirus24> Cool
14:27:12 <sigmavirus24> I appreciate that personally. My brain shuts off after I get past about 600 lines
14:27:20 <sigmavirus24> 400 means I can review something else =P
14:27:44 <mfedosin> yeah, we knew that
14:27:47 <ativelkov> You can always ignore tests, they are usually boring )
14:28:10 <sigmavirus24> nah, tests are important too
14:28:58 <mfedosin> so after I'm going to concentrate on our architecture and get rid of domain model. and work on stabilize API too. It means that import/export is not my highest priority
14:29:34 <ativelkov> Good. And I'll try to prepare a document for API-WG this week, so we may start discussing API-related work
14:29:47 <mfedosin> ativelkov, sg!
14:29:52 <ativelkov> We are out of time, BTW
14:30:11 <ativelkov> So, if there are open questions, let's continue in glance channel
14:30:19 <ativelkov> thanks all
14:30:20 <mfedosin> thanks, folks! and please review client commit :)
14:30:20 <ativelkov> #endmeeting