The W3C JSON-LD Community Group

JSON-LD Syntax and API

Go Back


W3C Logo

JSON-LD CG/WG — Minutes (Draft)

Date: 2022-06-22

See also the Agenda and the IRC Log

Attendees

Present: Ted Thibodeau Jr., David I. Lehn, Pierre-Antoine Champin, Melinda Minch, Roberto Polli, Orie Steel, Niklas Lindström, Eric Prud'hommeaux, Chris Mungall

Regrets:

Guests:

Chair: Gregg Kellogg

Scribe(s): Orie Steel, Gregg Kellogg, Pierre-Antoine Champin

Content:


1. Announcements and Introductions.

Gregg Kellogg: the CG as been anointed to make updated to specs / publish new ones… by the JSON-LD WG..
… to start off, we have had a lot of interest in YAML-LD lately..
… I'm Editor of JSON-LD 1.0 / 1.1, I've worked in CGs as well..

Niklas Lindström: I was also in the WG / co-author JSON-LD 1.0… I work a the national library of sweden, and we use jsonld.
… i've been interested in RDF * / JSON-lD *.

Anatoly Scherbakov: I work at datafold, i'm a software developer, interested in semantic data… I'm building an open source project that works on knowledge graphs.
… I first learned of these 10 years ago, but only now I have a project to work on them, excited to be here..

David I. Lehn: I work at Digital Bazaar, we've used JSON-LD for some time, current maintainer of jsonld.js… want to keep pushing the tech forward..

Ted Thibodeau Jr.: I'm with OpenLink Software, I'm in all W3C groups..

Philippe Le Hégaret: I'm the team contact for W3C for the JSON-LD WG..

Roberto Polli: I work for video <inaudible> department of italy, interested in web apis…. interested in standardization… I'm also interested in standardizing… .

Orie Steel: Pierre-Antoine: I joined the W3C Team, I've helped shepherd the RDF WG /RDF … interested in JSON-LD… and interested in augmenting it to cover RDF.

Melinda Minch: I'm based in BC, CA, I work on the World Data System, international technology office, I was given a task to build federated search for arctic research data… I'm interested in research dataset markup..

Orie Steel: I'm Orie Steel, CTO and founder of transumte. I'm an other of the DID spec and work with Verifiable Credentials..
… We apply JSON-LD DIDs for securing global trade documents for import/export..
… I'm interested in YAML-LD and YAML as a more human readable way of working with Linked Data..

Gregg Kellogg: to repeat for those who are not present yet, we are using IRC.

Melinda Minch: melinda-minch has joined #json-ld.

Gregg Kellogg: we use IRC for voting / recording minutes, etc..

Roberto Polli: Hi folks!.

Gregg Kellogg: You can use a native IRC client if you like..

Gregg Kellogg: https://www.w3.org/groups/cg/json-ld/calendar.

Gregg Kellogg: We have a F2F tentative in Vancouver.

Melinda Minch: like this? @present+ Melinda Minch.

Gregg Kellogg: see the calendar.

2. YAML-LD.

Gregg Kellogg: https://github.com/json-ld/yaml-ld/projects/1.

Gregg Kellogg: in the WG, we used the github projects feature to manage the open issues.
… we have created a number of issues.
… focusing first on use cases.
… eventually we will look at spec / testing.
… there is one issue regarding the spec and media types.

Gregg Kellogg: https://github.com/json-ld/yaml-ld.

Gregg Kellogg: the purpose of working issues, is to create use case documents or working towards a specification.
… typically we use pull requests, which means we will need to identify editors for each document managed in github.
… we are seeking editors.
… there is work required to transform use cases from issues.
… for example "compatibility with existing libraries".
… "as a developer I would like to use existing libraries so that work developing is easier…etc".
… we use "ReSpec" to document specs / use cases as HTML..
… we have a PR preview tool which helps visualizes the changes made.
… this allows for the community to review changes and make constructive comments.

Roberto Polli: Orie: to ease collaboration, it would be great to use bikeshed.

Gregg Kellogg: rather than trying to do issue review in this meeting, I would like to move on to other issues.

Roberto Polli: for building the document out of markdown.

Gregg Kellogg: any questions or comments before we move on/.

Roberto Polli: we will not discussion issues?.

Gregg Kellogg: I want to cover the agenda before getting to specific issues.
… and establish a regular call schedule.
… we may want to create different calls to cover other items.

Niklas Lindström: +1 for different calls for different topics.

Gregg Kellogg: for example JSON-LD *.
… there are also specification errata which needs to be addressed.
… the last thing I want to do is schedule our next call, and agree to recurring meetings.

Pierre-Antoine Champin: I wanted to comment on bikeshed vs respec.
… from what I know, bikeshed is a markdown version of respec.
… if folks want to use markdown, i think respec may support it now.

Roberto Polli: whatever you prefer, as long as it's text :).

Orie Steel: I've used both, and I think bikeshed works well..
… It can be hard to use if it's not set up properly, and there's some additional chances for errors or risks..
… I think that bikeshed is worth the lift if the advanced featuers warrant..

Gregg Kellogg: group has always used ReSpec… we have plugins / tools in ReSpec.
… AFAIK, every other RDF group has used ReSpec.
… my guess is that using ReSpec is probably an easier entry point.

Proposed resolution: Use ReSpec for spec and UCR documents.. (Gregg Kellogg)

Orie Steel: +1.

Pierre-Antoine Champin: +1.

Ted Thibodeau Jr.: +1.

Anatoly Scherbakov: +1.

Philippe Le Hégaret: 0.

Melinda Minch: 0.

Niklas Lindström: +0 (no strong opinion).

Roberto Polli: Can respec use .md ?.

Chris Mungall: sorry i joined late, should I be looking at something?.

David I. Lehn: +1, but i had never heard of bikeshed until just now, so not an informed vote.

Roberto Polli: I just want to avoid diffing .xml docs.

Roberto Polli: on PRs.

Eric Prud'hommeaux: +0.

Gregg Kellogg: ok so, last call for votes.

Resolution #1: Use ReSpec for spec and UCR documents..

Gregg Kellogg: we will use respec.
… any other discussion regarding YAML-LD ?.

3. JSON-LD-star.

Gregg Kellogg: https://github.com/json-ld/json-ld-star.

Gregg Kellogg: its fairly mature spec.

Gregg Kellogg: https://github.com/json-ld/json-ld-star/issues.

Gregg Kellogg: see the open issues, we hope to move it to a WG.

Pierre-Antoine Champin: RDF-star is an extension of syntax and semantics to use triples as the subject or object of other triples.

Gregg Kellogg: Draft JSON-LD-star spec: https://json-ld.github.io/json-ld-star/.

Pierre-Antoine Champin: this extension requires concrete syntax extensions, its been done for turtle… doing this for JSON-LD would cover this same update for JSON-LD.

eric: is there a playground for JSON-LD star.

Gregg Kellogg: I think there is only 1 implementation of JSON-LD start so far.

Gregg Kellogg: http://rdf.greggkellogg.net/distiller?command=serialize&format=jsonld&output_format=jsonld.

Gregg Kellogg: in the ruby version there is an option to pickup jsonld.
… and serialize as JSON-LD *.

Niklas Lindström: https://niklasl.github.io/ldtr/demo/?url=../test/data/diff-annotated.trig.

Niklas Lindström: I wanted to share a tool ^ right now its a simple demo of parse & serialize, but it has some visualization.
… I have added turtle-start, trig-star and jsonld-star.
… you can use the tool to explore the formats.
… its a pseudo playgroun.

eric: the bottom right corner ha a '<<' button which brings up an editor.

Gregg Kellogg: would you consider a PR to the JSON-LD repo to include a link?.

Niklas Lindström: I will look into that.

4. JSON-LD Spec updates.

Gregg Kellogg: there are a number of proposed updates to JSON-LD spec and Errata.
… this is important work for the group to get too.
… pchampin and I have been following these issues.
… plh, any info about the process to apply the errata.

Philippe Le Hégaret: the process was updated in Sep 2020, unfortunately, this was after JSON-LD was published.
… so the current JSON-LD Rec can not take advantage of the new process.
… because a recommendation have to "opt in" to the new process, and JSON-LD could not..
… The probability of the process evolving to make this retroactive is very low..

Gregg Kellogg: so when the group is ready to address this modification,.
… we will need to dedicate some time to dive into the new process..

5. Schedule next call.

Gregg Kellogg: I would propose that the group continues to use this timeslot, on a every-other-week basis.
… (with some exceptions during summer).
… every other week may not be enough in the future, but we can start with that..

Proposed resolution: go to every other week schedule in this timeslot, with the next meeting being July 6. (Gregg Kellogg)

Niklas Lindström: +1.

Gregg Kellogg: +1.

Pierre-Antoine Champin: +1.

David I. Lehn: +1.

Philippe Le Hégaret: 0.

Anatoly Scherbakov: +1.

Melinda Minch: 0.

Ted Thibodeau Jr.: +0.

Resolution #2: go to every other week schedule in this timeslot, with the next meeting being July 6.

Philippe Le Hégaret: will not be able to attend these calls on a regular basis,.
… but let me know if you need me..

Gregg Kellogg: ivan also said that he would try to attend, when possible..

Ted Thibodeau Jr.: can we put these calls in the JSON-LD WG/CG calendar?.

Gregg Kellogg: will do (already the case for this call).
… the minutes will also be included in the calendar, and validated during the next meeting.

Gregg Kellogg: https://github.com/json-ld/yaml-ld/projects/1.

Gregg Kellogg: anyone having an issue they want to start with?.

Gregg Kellogg: issue json-ld/yaml-ld#19.

ghurlbot: https://github.com/json-ld/yaml-ld/issues/19 -> Issue 19 Polyglot Modeling (VladimirAlexiev) UCR.

Gregg Kellogg: https://github.com/json-ld/yaml-ld/issues.

Eric Prud'hommeaux: how are people supposed to add UCRs?.

Gregg Kellogg: there is an issue template for UCRs in the https://github.com/json-ld/yaml-ld repo.

Eric Prud'hommeaux: (meta Q to see if we're on the same page).

Eric Prud'hommeaux: do people want to exceed JSON expressivity, or just use YAML as a nicer syntax for JSON?.

Gregg Kellogg: yaml-ld#8.

ghurlbot: https://github.com/json-ld/yaml-ld/issues/8 -> Issue 8 Round-trip safe json-ld -> yaml-ld -> json-ld (ioggstream) UCR.

Gregg Kellogg: I think there is an issue about that.
… (not the one above…).
… s/see/see/.

Gregg Kellogg: yaml-ld#17.

ghurlbot: https://github.com/json-ld/yaml-ld/issues/17 -> Issue 17 YAML-LD datatypes (and tags for datatypes) (VladimirAlexiev) UCR.

Orie Steel: +1 to JSON-LD and YAML-LD having a similar or identical set of capabilities.

Orie Steel: +1 to exploring both the "identical" and the "enhanced" paths.. at least initially.

Orie Steel: sorry I am talking now.

Orie Steel: I can take back over scribing ir you like.

Roberto Polli: we can start with a minimal set of feature.
… allowing to serialize only roundtrippable data (except for comments).
… once we have it, we can consider expanding it.
… for example, leveraging YAML anchors in order to reuse JSON bits in several places.
… (as long as they are not creating cycles).
… also, it is important to define a media-type.

Melinda Minch: Is it a goal to make these formats round-trippable between each other, or anything like that?.

Roberto Polli: ioggstream: q-.

Eric Prud'hommeaux: interestingly, YAML might no offer much referencing advantage (&node and *node) over JSON-LD 'cause it already had refs.

Orie Steel: roundtripping is important for interoperability.

Eric Prud'hommeaux: but not with the notion of copying data, more with re-using nodes.

Orie Steel: it is exciting to explore other features, but in a 2nd time.

Niklas Lindström: worried about the complexity we might open up.

Roberto Polli: ericP: refs and *node are different. One happens at json-ld level, the other one at seralization level..

Anatoly Scherbakov: I am uncertain how referencing can help; JSON-LD can assign an @id to a node, which can be used multiple times all over the document. Do we have to encourage a duplicate feature?.

Niklas Lindström: my own colleagues already have problem with the RDF part of JSON-LD, they merely see JSON.

Gregg Kellogg: thanks everyone, we have a lot to chew on.
… I encourage you to continue the discussion in the issues.
… From this discussion, I think we will start with a simple profile..

Orie Steel: thanks all!.

Anatoly Scherbakov: Thank you all very much!.

Melinda Minch: Thanks everyone, nice to meet you all.

Gregg Kellogg: rssagent, bye.


6. Resolutions

  • Resolution #1: Use ReSpec for spec and UCR documents..
  • Resolution #2: go to every other week schedule in this timeslot, with the next meeting being July 6.