DID WG Telco — Minutes
Date: 2021-01-26
See also the Agenda and the IRC Log
Attendees
Present: Brent Zundel, Ted Thibodeau Jr., Shigeya Suzuki, Joe Andrieu, Justin Richer, Manu Sporny, Adrian Gropper, Jonathan Holt, Drummond Reed, Daniel Buchner, Orie Steele, Kyle Den Hartog, Daniel Burnett, Geun-Hyung Kim, Juan Caballero, Kaliya Young, Brent Zundel
Regrets:
Guests:
Chair: Daniel Burnett
Scribe(s): Daniel Buchner
Content:
Daniel Buchner: Manu, is that clean PR good?
Manu Sporny: dbuc, haven’t had a chance to look at it yet – can do that now
1. PR Deadline
Brent Zundel: Sent out a notice a while back, you all should now have it
… the PR deadline is Feb 9
… Any PRs not raised by that day will be deferred to vNext
2. Special topic call
Brent Zundel: Will be this Thursday
… Purpose of the call is for anyone writing PRs to get feedback, guidance, aid, etc.
… Noon EST, on Thurs
Drummond Reed: I am planning to have the revised Appendix (based on the current comments and any discussion today) ready to review on the special topic call on Thursday.
3. Various WG Notes
Brent Zundel: Hoping to do a quick roundtable about some of the Notes that are being written
… Use cases is very nearly completely complete
Joe Andrieu: on the rubric: Daniel Hardman and I are meeting each week to push it forward
… Leveraging DHS/SVIP work
… Rubric proposed by that group as a starting point
… Will take the feedback in incorporate it into the Rubric
Brent Zundel: Test suite is the other major item
Orie Steele: The test suite has been refactored.
… Open PR in test suite repo. Open issue as to whether the test suite is about DID Method conformance or other variety of conformance
Orie Steele: https://github.com/w3c/did-test-suite/pull/19
Manu Sporny: a quick question: about DID Key, there may be multiple companies submitting implementations
… How should that work?
Orie Steele: no solution, but propose two tracks: everyone implements same features for a given method
… or track 2: different implementations, but limited to implementation tests, which may be less strict
Ted Thibodeau Jr.: implementation should be focus of test suite; that will deliver on method conformance/feature-coverage as a bonus
Ted Thibodeau Jr.: method as focus will not deliver on implementation conformance
Manu Sporny: Ted said it in IRC - thinks it is going to be problematic if we don’t focus on implementation-wide focus for testing
… With DID Key we’re seeing disjoint implementations, and the design decisions have been very different
Juan Caballero: @JoeAndrieu - I made a thingy for the DID-Rubric to help out Eric Welton at the DIF F2F last week - not sure if it’s useful but ping me when the cohort gets around to discussing did methods again? https://github.com/bumblefudge/blotter/blob/main/did-rubric-strawmen.md
Manu Sporny: Let us know how hard it’s going to be to focus the testing on cross-implementation conformance
Orie Steele: can support different implementations, but when we try to add up the coverage across many implementations, the reports will be spotty, given the different support across method implementations
… will there be a single set of test vectors per implementations all implementations will be judged by?
Manu Sporny: typically no one true set of vectors for an implementation
… just looking for at least two implementers implemented a feature
Manu Sporny: +1, thank you Orie!
Orie Steele: we should adjust the test suite until it does what Manu said
Daniel Burnett: next topic is the Implementation Guide
Drummond Reed: I was assigned, but will not have time to do it
… I am busy vaccinating the world
Daniel Burnett: Notes in process become the priority as we move to CR
… Chairs are leaning toward no Implementation Guide
4. Issues
Manu Sporny: Good news is we have many PRs
… heads up: check your PRs regularly
… lolly lolly lolly get your PRs in
… Two questions on Appendix and Persistence PRs
Drummond Reed: Joe’s new PR on Persistence is good, and please review, over to Joe for details
Joe: Did see the comments, not sure how URNs apply here
… URNs don’t do the thing they were brought in for
Drummond Reed: think we need 1/2 paragraphs that have generations of specs around the issue of URNs and the relationship to them
… think we need 1/2 paragraphs around the issue of URNs and the relationship to them
Manu Sporny: please take a look at that PR
Drummond Reed: we should shoot for Thursday, and give it a go ahead
Manu Sporny: those are PRs 457 and 460
… going to PR about DagCBOR
Brent Zundel: A section of text was moved by Manu as part of an editorial cleanup. Wasn’t a violation, but we need to move the text back
Orie Steele: lets just see the PR, and object to the PR
Manu Sporny: will move the text back, but objections will likely result
Daniel Burnett: we will treat this as a new set of objections for discussion
Jonathan Holt: notes issues with DagCBOR, knows about need for more eyes-on, but mostly concerned with the overarching CBOR representations, and thinks it needs to be reflected in the spec
… pulled much of the text from the canonicalizaton text from other spec, by this is a good approximation of a deterministic representation
… this was my attempt to create one such implementation of the representation
Orie Steele: isn’t this entire debate about did+dag+cbor ? pretty sure did+dag+cbor != did+cbor != did+ld+cbor
Jonathan Holt: I think we can get to some synergy where this is hardened and secure for the CBOR representations
Daniel Burnett: let’s not talk about At-Risk
Manu Sporny: let’s have that discussion on Thursday
… on CBOR canonicalization, DagCBOR
… background is in Issue 551
Orie Steele: See https://github.com/w3c/did-core/issues/551
Jonathan Holt: in absence of Jim Shad and Carsten, I am not an expert, but this was an attempt to get to an approach that would work
Daniel Burnett: At-Risk is marked because there is a concern that there may be insufficient implementations for a given feature
… risk of not having two implementations is the one and only thing that At-Risk is meant to convey
… it is an editor/chair decision
Jonathan Holt: What about the testable risk?
… I always see things being testable, not testable
Daniel Burnett: Manu, you take a shot at explaining testability and how we handle it
Manu Sporny: we have made a pass to identify testable statements. Some are clear, others are unclear how/if the WG can test things
… All tests should be in RFC language, and we need to look at ever normative statement, and ask ourselves, do we feel testable? Well, do we, punks?
… we expect normative statements to be testable by a machine
Daniel Burnett: if the editors feel something is not testable, you must throw down the gauntlet with a test
Jonathan Holt: if it’s one big number vs another big number, how to you handle errors in the test code/vectors?
Dan: we have an errata process, and if it’s a small bug, we correct it. We are more concerned with whether the issue is one that would occur outside of simple spec dev mistakes
Manu Sporny: we need to do this cleanly, else our charter time could run out, and we could get canned by the W3C
… ideally we should have multiple conforming implementations now, before CR
Orie Steele: as i was updating some of the tests, I was looking around for all the testable statements
… is there some tool or process to help divvy up work?
Manu Sporny: we don’t have a plan, but Amy has a normative statement extraction tool
… we will set aside a special call to go through those
Ted Thibodeau Jr.: GH has a feature called a Draft PR, could be of use for this
Kyle Den Hartog: can we use Amy’s tool and work on them now, or do we feel they will change?
Manu Sporny: brace yourself for change, CR-winter is coming
Daniel Burnett: we will start requiring a test if you submit a PR that could affect a test, and we use this as a natural PRDDoS mechanism
Manu Sporny: Orie, I know you resubmitted revocation/rotation PRs, how good do you feel about them?
Orie Steele: meeting with D. Hardman, and we’re getting there
… section on revocation is something everyone should get eyes-on. I am surprised by the position he takes
Orie Steele: See the PR on revocation https://github.com/w3c/did-core/pull/570
Manu Sporny: any issues you concerned about not being resolved?
Daniel Buchner: I have a concern: relative paths. if controller is a blank string.
… Is it a DID URI? have to walk back and forth. Seems hazy.
Orie Steele: https://github.com/w3c/did-core/issues/337
Orie Steele: relative pathing is confusing
… open issues exist about derefing, frags, base, etc.
… if you care about how these are handled in the spec, please look at Issue 337
… there are multiple ways relative refs can be formed
… Shares the concern about relative pathing
… please contributed to those examples if you want to de-FUDify
Shigeya Suzuki: I share similar feeling on relative path. it’s also to related to the relationship of resolver and method, I think.
Joe Andrieu: I still have on my plate to read through 100 pages regarding herd privacy, and I am hoping to get through it
Manu Sporny: we need to make resolution more concrete, and bring more specifics into our scope
… We can no longer operate on the notion that the first rule of Resolution Club is that we don’t talk about Resolution Club
Daniel Burnett: anything else for this wonderful Tuesday?