22:01:55 #startmeeting 2018-01-10 discussion 22:01:55 Meeting started Wed Jan 10 22:01:55 2018 UTC. The chair is wking. Information about MeetBot at http://wiki.debian.org/MeetBot. 22:01:55 Useful Commands: #action #agreed #help #info #idea #link #topic. 22:01:55 The meeting name has been set to '2018_01_10_discussion' 22:02:34 mrunalp, vbatts, stevvooe, tianon, crosbymichael, etc.? 22:03:16 [caniszczyk, Open Container Initiative] only major issue I'd like to bring up is OCI TOB election nomination schedule: https://github.com/opencontainers/tob/issues/33 22:03:20 oh man 22:03:30 hey 22:03:35 i'll dial in once i'm in the car 22:03:38 :-\ 22:21:22 #action wking to file a comment with a link to this log and a call for mrunalp and crosbymichael to weigh in 22:21:38 vbatts|work: other topics? 22:21:44 #topic distribution API 22:22:13 vbatts|work: not much progress last week. I began drafting an email covering the KubeCon notes 22:22:37 [caniszczyk, Open Container Initiative] here's a relevant github issue for distribution spec discussion: https://github.com/opencontainers/tob/issues/34 22:22:42 vbatts|work: we want to bring over the distribution API into OCI 22:22:49 [caniszczyk, Open Container Initiative] here's the gdoc vincent mentioned https://docs.google.com/document/d/15y0SBrrDFIEM7pnU-Oe3Y6pq-eTZfo0mk-k33cS2hR4/edit#heading=h.kmkpd9i5mwli 22:23:07 vbatts|work: I expect we want to get it circulated for discussion/TOB-vote in the next week or so 22:23:32 vbatts|work: but this is somewhat complicated by the TOB vote. There's no real reason to wait on the initial document 22:23:51 vbatts|work: but if discussion takes too long, we might need to start the discussion over with the new TOB 22:24:18 vbatts|work: I was thinking to not wait 22:24:24 stevvooe: I was thinking we would wait 22:24:34 stevvooe: when does the new TOB come online? 22:24:40 cracra: January 29th 22:24:48 vbatts|work: all the voting before then? 22:25:10 cracra: nominations finish next Thursday. Votes close on the 26th, and the change happens on the 29th 22:25:18 vbatts|work: is there an overlap period? 22:25:28 cracra: I think votes stop counting when your term is up 22:25:40 cracra: so if the vote comes in before the 29th it counts 22:26:11 I think it would be clearer to have a vote that did not span the 29th 22:26:33 cracra: this project doesn't seem as complex as some of our previous ones 22:26:56 cracra: the formal process requires a TOB pull request and formally asking the TOB for a vote 22:27:27 vbatts|work: I think it's fine to proceed, and we'll figure out when the vote falls later 22:27:47 cracra: a finallized proposal by the 19th would allow plenty of time for a vote 22:28:12 stevvooe: I'd prefer to wait for the new TOB instead of pushing it through 22:28:30 cracra: that's fine. I think we can start working on the proposal and see when it's ready 22:28:52 stevvooe: I don't think the active TOB will affect the proposal 22:29:13 cracra: I'll take that as an action item that everyone will collaborate on that Google Doc 22:29:17 vbatts|work: sounds good 22:30:07 #topic open PR for VMs in runtime-spec 22:30:23 [caniszczyk, Open Container Initiative] !link distribution spec https://github.com/opencontainers/tob/issues/33 22:30:23 #link https://github.com/opencontainers/runtime-spec/pull/405 22:30:37 [caniszczyk, Open Container Initiative] #link distribution spec https://github.com/opencontainers/tob/issues/33 22:30:43 vbatts|work: to my memory, this was punted to see how thin-VM containers shake out 22:30:48 [caniszczyk, Open Container Initiative] i'm such a meetbot noob 22:31:28 vbatts|work: I think now we are at the point where thin-VM containers are clearer 22:31:33 stevvooe: do we know what's missing? 22:31:53 vbatts|work: we may need to start the discussion over because things may have changed since the previous discussion 22:32:02 stevvooe: the runtime-spec PR seemed pretty minimal 22:32:32 vbatts|work: runV can support multiple backends, so what is the minimum viable set of properties for thin-VM runtimes? 22:33:11 vbatts|work: I have nothing else right now 22:34:24 stevvooe: there was some discussion of JSON Schema on the mailing list... 22:34:30 stevvooe: but we can all just leave... 22:34:51 #topic JSON Schema 22:35:24 stevvooe: I just keep running into bugs. I'm not sure if it's JSON Schema or gojsonschema. But when I went to modify it, it took me half a day to add a field 22:35:32 stevvooe: in containerd we've had to add our own validation 22:35:59 vbatts|work: I don't know what ripping it out would... 22:36:18 stevvooe: It's currently causing the runc build to fail, because runc doesn't pin gojsonschema 22:36:25 vbatts|work: so this is a Golang issue? 22:37:12 stevvooe: I'm concerned that folks will feel the JSON Schema covers the spec, because it's only really covering a subset 22:37:41 #link https://github.com/opencontainers/runc/issues/1680 22:38:29 stevvooe: I agree with the compact-ness sentiment, and I'm a huge fan of protobuf and similar for schema declaration 22:38:35 stevvooe: but JSON Schema reminds me of XML 22:38:41 vbatts|work: XML was better at this 22:38:48 stevvooe: XML was way better at this 22:39:07 #link https://github.com/opencontainers/runtime-spec/pull/945 22:39:36 vbatts|work: I think we already have docs saying the Markdown is normative, and the JSON Schema is informative 22:39:53 #link https://github.com/opencontainers/image-spec/pull/739 22:40:32 #link https://github.com/opencontainers/runtime-spec/blame/v1.0.1/config.md#L6 22:40:41 ^ current runtime-spec example showing that the JSON Schema is informative 22:41:17 #endmeeting