21:01:39 <wking> #startmeeting 2016-10-26 discussion
21:01:39 <collabot> Meeting started Wed Oct 26 21:01:39 2016 UTC.  The chair is wking. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:01:39 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
21:01:39 <collabot> The meeting name has been set to '2016_10_26_discussion'
21:01:47 <wking> #chair mrunalp vbatts
21:01:47 <collabot> Current chairs: mrunalp vbatts wking
21:02:09 <mrunalp> jlb13, hey!
21:02:24 <jlb13> hey there
21:02:28 <philips> mrunalp: in!
21:02:32 <jlb13> i'm woefully behind on... everything. but, at least present
21:03:34 <wking> #topic image update
21:03:55 <wking> philips: the rc2 vote failed due to abstains.  I'll just repost the same vote
21:04:12 <wking> philips: there are a few items out for rc3.  I'm not sure if those got completed...
21:04:38 <wking> stevvooe: I had some stuff about package organization.  I merged wking's PR; do we need to do anything else?
21:04:55 <wking> philips: we need to explicitly explain the split to avoid repeating the conversation in the future
21:05:02 <wking> philips: there was also discussion about signatures
21:05:06 <wking> stevvooe: I opened a new issue
21:05:20 <philips> https://github.com/opencontainers/image-spec/issues/400
21:05:22 <wking> #link https://github.com/opencontainers/image-spec/issues/400
21:05:24 <philips> that is the issue
21:05:39 <wking> stevvooe: I've had this conversation a few times before, and tried to lay out the common blockers and such
21:05:48 <wking> philips: otherwise I'll leave it to vbatts and -rc3
21:06:16 <wking> vbatts: this past week I realized that the runtime doesn't have a runtime rc3.  I don't even know if the image-spec has an rc3 milestone
21:06:20 <philips> there is an rc3 milestone https://github.com/opencontainers/image-spec/milestone/10
21:06:23 <philips> for image-space
21:06:25 <philips> for image-spec
21:06:26 <wking> vbatts: we should start sorting issues and PRs into those milestones
21:06:31 <stevvooe> rc3 link for image-spec https://github.com/opencontainers/image-spec/milestone/10
21:06:39 <wking> vbatts: if you want something in those milestones, let us know
21:07:08 <stevvooe> also, opened https://github.com/opencontainers/image-spec/issues/415, on windows layer application
21:07:14 <wking> #link https://github.com/opencontainers/runtime-spec/milestone/12
21:07:56 <wking> vbatts: stevvooe had wanted a design doc for...
21:08:14 <wking> stevvooe: I see a lot of scattered PRs, and the high-level discussion gets lost in the weeds of those PRs
21:08:31 <wking> +1 to having a plan :)
21:09:47 <wking> vbatts: it would be helpful to have spec anchors so we can link to MUST/SHOULD/etc. requirements
21:10:25 <wking> vbatts: It would be awesome to have stable URLs for each requirement so validation tools can link you to the rule you're breaking
21:10:39 <RobDolinMS> (IRC only) I believe writing the spec in MarkDown gives us anchors in .MD.  I can check if these get output in the HTML
21:10:57 <wking> RobDolinMS: headers give anchors, but not every sentence
21:11:14 <RobDolinMS> Good point wking
21:11:17 <wking> vbatts: sometimes there are multiple requirements in a single sentence.  Avoiding that might be a style issue
21:11:31 <wking> stevvooe: we'll also need multiple tests per requirement
21:11:37 <wking> vbatts: agreed
21:12:00 <wking> stevvooe: so do we want a tagging format?
21:12:29 <wking> You can dump HTML anchors in wherever you want
21:13:09 <wking> vbatts: we need tests for every MUST, and when validating we should be able to link to the spec for that MUST
21:13:28 <wking> duglin: Are you saying we say "MUST" where we don't mean MUST in the RFC sense?
21:13:31 <wking> vbatts: no
21:13:39 <stevvooe> something like "<required name="Manifest-20"></required>"
21:13:54 <wking> vbatts: and instead of failing hard, we should be printing as many failures as we can find
21:14:02 <wking> stevvooe: lets open an issue around this
21:14:06 <RobDolinMS> @vbatts - I'm glad to help with this one
21:14:06 <collabot> RobDolinMS: Error: "vbatts" is not a valid command.
21:14:15 <RobDolinMS> vbatts - I'm glad to help with this one.
21:14:29 <RobDolinMS> We did an index for https://github.com/microsoft/api-guidelines/ and may be able to borrow from some of this
21:14:33 <wking> stevvooe: the REST compiler has a list of errors it can output, we can use that for prior art
21:14:46 <stevvooe> s/REST/RUST
21:14:58 <wking> ^thanks
21:15:15 <stevvooe> https://doc.rust-lang.org/error-index.html
21:15:17 <RobDolinMS> Per vbatts: Output should include:
21:15:47 <wking> #link https://github.com/opencontainers/runtime-tools/issues/16
21:15:51 <RobDolinMS> Test case | JSON snipped tested | reference to relevant spec
21:15:53 <wking> runtime-tools issue about this^^
21:16:54 <wking> vbatts: I've been working though tools that will interact with the image spec to kick its tires
21:17:10 <wking> vbatts: the more tire-kicking we can get, the better
21:17:14 <mrunalp> https://github.com/opencontainers/runtime-spec/pull/518/files
21:17:21 <wking> #topic runtime-spec process.terminal handling
21:17:26 <wking> #link https://github.com/opencontainers/runtime-spec/pull/518
21:17:54 <wking> mrunalp: what if we don't say how /dev/console is setup, but we tie its existence to process.terminal
21:18:36 <wking> I'm fine landing #518 and then opening a separate issue about splitting /dev/console existence from process.terminal
21:18:44 <wking> mrunalp: then it just needs another LGTM
21:18:48 <mrunalp> https://github.com/opencontainers/runtime-spec/pull/513
21:18:54 <wking> #topic runtime-spec command line API
21:18:59 <wking> #link https://github.com/opencontainers/runtime-spec/pull/513
21:19:10 <wking> mrunalp: this looks fine to me.  It needs more review from maintainers
21:20:00 <wking> I'm not aware of any other major issues.  I'd like to land #513 and tune in follow-up PRs
21:20:12 <wking> mrunalp: anyone else have thoughts on #513?
21:20:16 <wking> mrunalp: any other runtime topics?
21:20:45 <wking> philips: in the remaining 20 minutes, vote on the -rc2 email ;).
21:20:49 <wking> mrunalp: and review #513 ;)
21:20:53 <wking> #endmeeting