JSON-LD WG — Minutes

Date: 2023-12-13

See also the Agenda and the IRC Log


Present: TallTed, David I. Lehn, Benjamin Young, Gregg Kellogg, Dave Longley, Anatoly Scherbakov, Pierre-Antoine Champin, Niklas Lindström



Chair: Benjamin Young

Scribe(s): Anatoly Scherbakov, Pierre-Antoine Champin, Gregg Kellogg


Benjamin Young: we are going to discuss the future of the WG, future publications and the road ahead.

1. Announcements and Introductions.

Benjamin Young: Any announcements or introductions?

Gregg Kellogg: YAML-LD Final Report was published yesterday. Posted on CG blog on it.

Manu Sporny: hello and greetings!

Dave Longley: great to see everyone!

2. Updates from the CG.

Benjamin Young: any further updates from the CG for Dave and Manu?

Gregg Kellogg: CG was working on YAML-LD, and driving JSON-LD issues.
… Latest ones revolve around @nest and scoped contexts.
… We introduced CBOR-LD a few weeks ago and discussed it a bit last week. Looking forward to hear about DigitalBazaar’s work on that.

Benjamin Young: thanks! A lot of work has accumulated to the point that we’re looking into formalizing it.

Pierre-Antoine Champin: I discussed with colleagues about Linked Data in WoT scenarios; they’re interested in CBOR-LD and might join the WG.

Niklas Lindström: we’ve been talking about JSON-LD-Star and RDF-Star and thinking how to integrate these.

3. Plans/desire to publish Best Practices doc, YAML-LD, and a CBOR-related specification.

Benjamin Young: Let’s start with low-hanging fruit.

3.1. Best Practices.

Gregg Kellogg: https://w3c.github.io/json-ld-bp/.

Benjamin Young: the question about Best Practices has been there for a while. Do we need to post it as a separate note?

Benjamin Young: current charter: https://www.w3.org/2023/01/json-ld-wg-charter.html.

Pierre-Antoine Champin: I agree; I need to check with the charter and not everyone might agree.

Gregg Kellogg: there was never a resolution to publish the BP as a Note. It shows the last published version and that leads to a 404. Might be a problem with ReSpec.
… we can certainly publish it as a Draft Note, even if it is incomplete.
… there is also a Best Practices document in the CG but it’s been removed from the UI.

Benjamin Young: do they differ?

Gregg Kellogg: they’re different.

Benjamin Young: maybe some resolving is in order. The Charter only spells out maintenance of JSON-LD normative documents and also allows non-normative documents.
… to make YAML-LD and CBOR-LD normative we need to move formally as a group to re-charter.

3.2. YAML-LD.

Pierre-Antoine Champin: anatoly-scherbakov: nice to meet Manu and Dave.

Benjamin Young: is it ok for the group to be named JSON-LD if we extend the scope to YAML-LD, CBOR-LD… ?
… Should we find an Umbrella term?

Manu Sporny: I agree with Anatoly, we should shift the WG name given the expanded scope.

Dave Longley: JSON-LD and others! WG :)? … JSON-LD and Derivatives WG (doesn’t sound as friendly).

Benjamin Young: Also other formats such as CSV, Parquet… could be addressed.

Manu Sporny: agree with Anatoly, makes sense to rename the group to focus on expanded scope.

Dave Longley: Expanded JSON-LD Universe WG.

Gregg Kellogg: there is a standard for CSV for LD, its ten years old and has moderate use.

Benjamin Young: https://www.w3.org/TR/tabular-data-primer/.

Gregg Kellogg: “tabular data on the web” it is. It probably needs to be revisited, needs periodic updates. I think though that CBOR is definitely inspired by JSON, YAML and JSON developed together.
… it might be confusing to try to come up with some other name (Linked Data Working Group? - that is too broad maybe).

Dave Longley: JSON-LD Umbrella WG.

Gregg Kellogg: but we can probably stick with the JSON-LD as group name as we’re working on things closely related to JSON-LD.

Benjamin Young: agree, and let’s move on though.

Niklas Lindström: I agree this is a tricky question; I am leaning towards what Gregg said. One the reasons of JSON-LD success is because it is RDF channeled through JSON.

Gregg Kellogg: CSV on the web came out before JSON-LD 1.0 was standardized.

Niklas Lindström: CSV on the Web saw lower adoption. Something about JSON is very useful, I do not know how to call it more abstractly so that it rings as well as it does now.
… there’s something in the simplicity of JSON-LD itself.

Gregg Kellogg: INFRA-LD.

Niklas Lindström: leaving JSON behind we miss the point how we got here.

Niklas Lindström: “Not-XML-LD”.

Dave Longley: JSON-LD and Friends.

Pierre-Antoine Champin: My own opinion: I agree Tabular Data on the Web would deserve a refresh. Having one group for all kinds of data formats wouldn’t be optimal though. We are focusing on JSON, one particular shape of data.

Manu Sporny: agree that we need to focus in the new WG.

Benjamin Young: +1.

Pierre-Antoine Champin: other concerns, other languages, should probably be addressed by other groups.
… JSON-LD WG should care about JSON and very similar formats.

Dave Longley: +1 that JSON-LD is the unifier / north star / commonality.

Benjamin Young: thanks everyone! Let’.
… thanks everyone! Let’s keep JSON in focus.

Gregg Kellogg: .

Gregg Kellogg: .

Benjamin Young: CG report for YAML-LD published, thanks Gregg! what’s the future of this format in the WG?

Dave Longley: +1 to YAML-LD going standards track.

Manu Sporny: I support YAML-LD to go to Standards Track, as long as someone can help moving it through the process.
… it provides a signal that we’re onto something, these patterns are useful in other syntaxes, it allows the RDF data model to shine.
… in the other syntaxes you can express the same data model: JSON-LD - YAML-LD - CBOR-LD and back, that’s a good thing.
… we should take this to Standards Track. What about implementations?

Pierre-Antoine Champin: anatoly-scherbakov: the first implementation of YAML-LD is probably gkellogg’s.

Manu Sporny: I’m developing a Python implementation, based on PYLD, alpha stage at this point.
… It passes the YAML-LD test suite. My next step is to run the JSON-LD test suite.
… I’m using it in a little project of mine: browser and knowledge workspace for LD, mostly based on YAML-LD.

Niklas Lindström: this gets us two baseline implementations. Moving through WG will be about that. Any other notices about implementations?


Gregg Kellogg: https://w3c.github.io/json-ld-cbor/.

Benjamin Young: let’s move on to CBOR-LD. The questions about it are centered about progressing CBOR-LD spec to match implementations.

Gregg Kellogg: https://digitalbazaar.github.io/cbor-ld-spec/.

Benjamin Young: what level of compression should we use?

Manu Sporny: Introduction to CBOR-LD: https://lists.w3.org/Archives/Public/public-json-ld-wg/2020Jul/att-0004/Introduction_to_CBOR-LD.pdf.

Dave Longley: i believe there are three implementations, one in Rust, one in Java.

Manu Sporny: putting the link about CBOR-LD abortions. There are presentations from 2020 we published, they go over the basics. CBOR-LD has 2-3 implementations so far: JS, Rust and something else.

Dave Longley: one in JavaScript.

Manu Sporny: primary reason was Verifiable Credentials, we had a program in the US with the National Association of Convenience Stores about digital age verification program.
… the goal was privacy preserving age verification.
… so that we can prove your age without disclosing any other PII.
… that was in 2018-2019. One of the things they needed was ability to scan the verifiable credential which was a JSON-LD document.
… thus we needed a very high density bar code so that the old hardware can scan and handle it, we needed to get a JSON-LD document down to 350 bytes.
… that’s why CBOR-LD came into existence, we needed to compress digitally signed JSON-LD so that it could fit into a QR code. We’re now in production.
… about 4-6 months ago, California department of Motor Vehicles launched their digital driver’s license which includes a CBOR-LD QR code.
… in California you can now show that QR code which is CBOR-LD and prove your age.

Anatoly Scherbakov: anatoly-scherbakov: Well IMHO that’s really super cool.

Manu Sporny: the rollout is still happening but I wanted to make a point that it is already in production and in practical use.
… we put a version number on the version that’s out there so that W3C WG can introduce breaking changes in a new version.
… spec is not in a good shape, it is out of date from the implementation.

Dave Longley: -q.

Manu Sporny: we’ve talked about the plan to merge the changes in current spec and the reality of implementations.
… we have an uncompressed mode in the spec. Even that saves a number of bytes, but compression is what the real users are interested in.
… we also are working with governments about integrating CBOR-LD into their digital ID systems.
… that’s kind of where we are with CBOR-LD.

Gregg Kellogg: I think the two different documents address different things. Pierre-Antoine’s expresses basic JSON with CBOR compatible with JSON-LD. DigitalBazaar’s version is mostly about how you provide that with semantic compaction.
… YAML-LD sets a pattern we probably want to stick with — it is mostly API centric.
… it mostly involves transformation between YAML and JSON-LD Internal Representation. Compression doesn’t happen in CBOR, it can happen in Internal Representation.
… A concern I had: CBOR-LD 1.0 version doesn’t have a parallel in core CBOR.

Pierre-Antoine Champin: Gregg summarized this very well.

Dave Longley: about magic numbers: current implementations have tags indicating their CBOR-LD and version numbers.
… implementations support that; not sure if spec reflects it.

Gregg Kellogg: the current spec doesn’t detail that. Not sure how the tag structure with JSON-LD in it extracted from a CBOR document is distinguished against any other CBOR structure.

Manu Sporny: CBOR has a registry, tags are registered there, what we need is to request new numbers in the registry which are granted on first come first serve basis. You don’t need an official structure to claim them.
… if we’re an official WG it makes it even easier to register our signature bytes in the CBOR tags registry.

Benjamin Young: do we want to move forward as chartered, keeping YAML-LD as a Note or a Draft Note, and bringing CBOR-LD spec to the status of a Note?
… or we feel we are ready to bring these to Standards Track sooner and recharter the WG at this point?

3.4. rechartering?

Gregg Kellogg: I think there is sufficient implementation done for both to move to the Recommendation Track.

Dave Longley: +1 to gregg.

Gregg Kellogg: I do not know if publishing CBOR-LD as a Note makes a difference. Bringing it to Rec Track will improve visibility and hopefully drive participation.
… Updated Charter should also help other things, like fostering JSON-LD specs.
… Charter doesn’t specifically need to mention RDF-Star.

Benjamin Young: currently the charter only calls for maintenance, i.e. non breaking changes.

Pierre-Antoine Champin: errata is appropriate; not sure how breaking a change might be to fix a bug.
… the Note track and the Rec track are meant for different kinds of documents.
… it would be a pushback if we post a spec as a Note and then move it to Rec.
… it is Rec material. Falling back to Note track if we are not allowed to push it to Rec wouldn’t work. It is a document meant to be a Recommendation, not a Best Practices, not a Note.
… even if we could, having it on Note track can do it better than a CG report.
… we will continue working on it in the CG of course until the WG is allowed to take it to Rec track.

Manu Sporny: we want to move CBOR-LD out of digiatlbazaar github repo and move it to JSON-LD Github repo.
… can we do that? secondly, CBOR-LD is way behind where YAML-LD spec is. Problem is, many of us are heads down in Verifiable Credentials WG trying to get about five specs to Rec track.
… we do not have much bandwidth to work on this spec. Maybe it becomes easier next summer.
… we’re discussing going into production with CBOR-LD systems with national and state governments. They don’t necessarily want to wait until the standard is done.
… but they want an acknowledgement of W3C that W3C is looking forward to standardize CBOR-LD.
… CBOR-LD as a Note doesn’t make a lot of sense, we’d like it to go to Rec track.
… a new WG charter mentioning CBOR-LD could be a signal to the governments and buy us a bit of time to get the spec into proper shape.
… YAML-LD is further along and we could recharter the group putting both in scope, and focus on YAML-LD first then switching to CBOR-LD later.
… and we only have 6 months to get CBOR-LD done.

Pierre-Antoine Champin: +1.

Manu Sporny: publishing a new charter would be a positive signal to the community that we’re working on these.

Niklas Lindström: +1.

Niklas Lindström: sounds like a good idea. I think what we define should be very minimalistic, kind of glue code defining the serialization.
… this could also say that JSON-LD is beyond JSON. Contextual Compaction of Linked Data with a kind of Framing is the overarching theme here.

Benjamin Young: suggesting we take an action today to bring CBOR-LD into CG space.
… let’s start its life there, it will need much discussion and activity.

Gregg Kellogg: nominally it’s a CG action but we are highly overlapped and we can resolve to do that.

Manu Sporny: agree, gkellogg – we should write all of the concerns / issues we have down into the issue tracker.

Gregg Kellogg: manu mentioned that there are open issues with the spec. Would be great if github issues reflect those. this will make it easier for people to contribute.

Proposed resolution: Bring Digital Bazaar’s CBOR-LD 1.0 editor’s draft https://digitalbazaar.github.io/cbor-ld-spec/ into the JSON-LD CG for future work. (Benjamin Young)

Manu Sporny: +1.

Anatoly Scherbakov: +1.

Gregg Kellogg: +1.

Dave Longley: +1.

TallTed: +1.

Niklas Lindström: +1.

Pierre-Antoine Champin: +1.

Benjamin Young: +1.

David I. Lehn: +1.

Resolution #1: Bring Digital Bazaar’s CBOR-LD 1.0 editor’s draft https://digitalbazaar.github.io/cbor-ld-spec/ into the JSON-LD CG for future work.

TallTed: technically, “adopt Digital Bazaar’s CBOR-LD”.

Benjamin Young: resolved. David you apparently hold the super powers, can you do the actual moving please?

David I. Lehn: eventually it will be moved to W3C and we will have to move it again?

Gregg Kellogg: yes if the WG is rechartered. We’ll move the repos from the CG to the WG github org, but this might take months.

David I. Lehn: this will mean broken links.

Benjamin Young: we can’t get it into W3C repo now because it is CG material.
… there is an ambient consensus about rechartering.

Gregg Kellogg: let’s do a proposal.

Proposed resolution: Recharter the JSON-LD WG to focus on YAML-LD and CBOR-LD. (Benjamin Young)

Manu Sporny: +1.

Anatoly Scherbakov: +1.

Dave Longley: +1.

Gregg Kellogg: +1.

Pierre-Antoine Champin: +1.

Benjamin Young: +1.

TallTed: +1.

David I. Lehn: +1.

Niklas Lindström: +1 (not excluding RDF-star alignment?).

Resolution #2: Recharter the JSON-LD WG to focus on YAML-LD and CBOR-LD.

Benjamin Young: this resolution doesn’t need to be exclusive, it just signals we want to recharter.
… we are still continuing the maintenance of JSON-LD core specs and other things.
… we’ll likely not have another CG call before the end of the year, we’ll get back to this in January.

Benjamin Young: Zakim: adjourn meeting.

Benjamin Young: Zakim: end meeting.

4. Resolutions