22:04:19 #startmeeting OCI 1/11 22:04:19 Meeting started Wed Jan 11 22:04:19 2017 UTC. The chair is mrunalp. Information about MeetBot at http://wiki.debian.org/MeetBot. 22:04:19 Useful Commands: #action #agreed #help #info #idea #link #topic. 22:04:19 The meeting name has been set to 'oci_1_11' 22:04:51 oh crap 22:05:05 hrm. i'm stuck in a meeting that is running long 22:05:14 let me try to sneak out 22:06:18 stevvooe, not heard from many maintainers 22:06:26 many PRs stuck on go-digest 22:08:07 mrunalp: chair me? 22:08:18 #chair wking 22:08:18 Current chairs: mrunalp wking 22:08:46 stevvooe: we should just have the maintainers chime in explicitly for new project proposals 22:08:56 mrunalp: anything else on this topic? 22:09:35 stevvooe: I need to get the admin items through. I have some image-spec PRs that are related 22:09:58 #link https://github.com/opencontainers/image-spec/pull/486 22:10:04 #link https://github.com/opencontainers/image-spec/pull/512 22:11:07 stevvooe: if we want to bring things into OCI and this is the experience (bureaucracy) it's not going to happen often 22:11:13 stevvooe: we need to make onboarding easier 22:11:39 RobDolinMS: do we want to review the PRs to see if they're signed? 22:11:49 mrunalp: we'll get to that topic in a bit 22:12:05 #topic many minor runtime-spec PRs 22:12:20 mrunalp: there are a lot of small PRs, and many of them need to be rebased 22:12:37 mrunalp: there are PullApprove issues with some of RobDolinMS's PRs 22:12:45 RobDolinMS: I'll get to those tomorrow afternoon 22:12:57 mrunalp: let us know if you need help on that. It may be a PullApprove issue 22:13:38 somebody?: Also check that GitHub has your Microsoft email 22:13:51 mrunalp: any particular PRs that need resolution? 22:14:17 mrunalp: do we want to traige them online? 22:14:28 RobDolinMS: I think that's useful to do as a collective 22:15:16 crosbymichael: we need to split out the CLI stuff, since they're blocking on each other 22:15:41 crosbymichael: we should be able to get a 1.0 without handling the CLI stuff 22:16:11 crosbymichael: we need to freeze runtime-spec structs 22:16:26 mrunalp: my issue with that would be that there would be no way to validate 1.0 compliance 22:16:55 crosbymichael: I don't mean release 1.0 without it. I mean, finalize the structs and JSON and then come back around to the CLI 22:17:33 mrunalp: so maybe send an email today about a feature-freeze on the spec (in a week or two), and then come back to the CLI 22:18:09 https://github.com/opencontainers/runtime-spec/pulls?page=1&q=no%3Amilestone+is%3Aopen&utf8=%E2%9C%93 22:18:50 crosbymichael: two milestones (one for the spec, and one for the CLI)? 22:18:56 mrunalp: the CLI is just the one PR now 22:19:21 RobDolinMS: cracra, can you give me privileges to create and assign to milestones 22:20:03 mrunalp: I see the shared screen now 22:20:28 https://github.com/opencontainers/image-spec/pull/512 at someone's leisure 22:21:34 RobDolinMS: newest to oldest, or oldest to newest? 22:21:40 mrunalp: newest to oldest 22:22:05 #link https://github.com/opencontainers/runtime-spec/pull/647 22:22:09 mrunalp: we can make that 1.0 22:23:29 I think we want it 22:23:37 #link https://github.com/opencontainers/runtime-spec/pull/646 22:24:30 mrunalp: added to 1.0 22:24:47 #link https://github.com/opencontainers/runtime-spec/pull/635 22:25:20 added to 1.0 22:25:37 #link https://github.com/opencontainers/runtime-spec/pull/630 22:25:52 mrunalp: I don't think this is ready 22:26:03 RobDolinMS: 1.1? 22:26:11 crosbymichael: this one is weird 22:26:27 crosbymichael: it's not cgroups at all 22:26:33 crosbymichael: it's more like an rlimit 22:26:42 crosbymichael: and this wouldn't be a breaking change if it landed later 22:27:02 RobDolinMS: added to 1.1 and pinged the OP about that 22:27:42 stevvooe: done 22:27:50 #link https://github.com/opencontainers/runtime-spec/pull/628 22:27:54 crosbymichael: why even do this^ 22:27:59 crosbymichael: because asciidoc rendered better 22:28:02 mrunalp: it makes adding anchors and things easier 22:29:07 crosbymichael: learning new syntax is a big hurdle 22:29:27 stevvooe: seconding that. Would be good to get what we need from Markdown 22:29:46 RobDolinMS: This is a problem in many orgs (e.g. with the W3C). We should see what they're doing 22:30:07 stevvooe: we changed something, which made it easier for a small number of people but then made it difficult for lots of people 22:30:16 stevvooe: we don't want to project pain to the rest of the community 22:30:28 crosbymichael: all the OCI projects should be using the same tooling 22:30:44 #link https://github.com/opencontainers/runtime-spec/pull/620 22:30:47 RobDolinMS: mikebrow ? 22:31:00 mikebrow: this is ready to go 22:31:37 #link https://github.com/opencontainers/runtime-spec/pull/611 22:32:19 mrunalp: any strong opinions? 22:32:46 RobDolinMS: stevvooe? 22:32:51 stevvooe: looking 22:33:29 stevvooe: This looks like it's the host-local runtime-config? 22:33:31 yes 22:33:39 stevvooe: Media types should be controlled by the image spec 22:34:29 stevvooe: it doesn't make a whole lot of sense for the runtime config to have a media type 22:34:43 this is just "should it have a media type" 22:35:12 #link https://github.com/opencontainers/runtime-spec/pull/611#issuecomment-261700992 22:35:50 stevvooe: the runtime config is not something you set and don't touch 22:36:35 stevvooe: let the triage keep going, and I'll think about this one 22:37:02 #link https://github.com/opencontainers/runtime-spec/pull/577 22:37:15 RobDolinMS: this is a one-worder 22:37:33 needs a rebase 22:39:01 #link https://github.com/opencontainers/runtime-spec/pull/570 22:39:08 I'll split that out into separate PRs^ 22:39:19 #link https://github.com/opencontainers/runtime-spec/pull/555 22:40:18 RobDolinMS: stevvooe, crosbymichael does legacy-Docker have a preference? 22:40:32 stevvooe: If you want host namespaces, don't set the field 22:40:39 oops, that was crosbymichael^ 22:40:54 mrunalp: agreed. If you don't need new namespaces, don't add the property 22:41:19 stevvooe: if you allow nulls you bring in tri-state logic, which can be hard across languages 22:41:45 RobDolinMS: so close without merging? mrunalp, can you do that? 22:41:48 mrunalp: ok 22:42:14 stevvooe: is the field optional? 22:42:36 mrunalp: don't we require a mount namespace? 22:42:38 crosbymichael: no 22:43:16 It's not marked required in schema/config-linux.json either 22:43:41 RobDolinMS: I'm shifting to another meeting. Ping me if you need me to rebase anything else 22:43:44 mrunalp: continue? 22:43:58 #link https://github.com/opencontainers/runtime-spec/pull/552 22:44:27 mrunalp: assign to vbatts? 22:44:51 #link https://github.com/opencontainers/runtime-spec/pull/548 22:45:10 crosbymichael: there should be no string pointers, right? Not ever 22:45:29 crosbymichael: strings can be optional, but you don't need pointers in the struct 22:46:05 https://github.com/opencontainers/runtime-spec/blob/master/specs-go/config.go#L295 22:46:08 https://dave.cheney.net/2013/01/19/what-is-the-zero-value-and-why-is-it-useful 22:46:54 mrunalp: so remove the existing string pointers? 22:47:18 mrunalp: I can make a PR for that 22:48:34 #link https://github.com/opencontainers/runtime-spec/pull/545 22:48:44 this may not make sense now that some -1 stuff has landed 22:48:51 mrunalp: I'll ping hqhq and we'll see 22:49:52 #link https://github.com/opencontainers/runtime-spec/pull/540 22:50:27 I still think this is useful. I'm fine with modifications, but if the spec is against them (and it seems to be), I'd rather have that a generic rule that's not tied to a particular feature set 22:52:08 mrunalp: the hostname MUST seems sufficient 22:52:24 it is, but then you have to annotate every possible field 22:53:40 you could have a spec that let you tweak existing resources 22:53:56 stevvooe: the spec could just warn about side-effects (instead of blocking) 22:54:22 so close this PR and file a new one to lift the no-tweaking restriction? 22:55:36 stevvooe: yeah. Remove that config-linux.md line 22:55:50 and there are other no-tweaking restrictions to lift as well. I'll file a PR 22:55:54 mrunalp: out of time? 22:55:56 crosbymichael: yup 22:56:00 #endmeeting