The W3C JSON-LD Community Group

Go Back

W3C Logo

JSON-LD CG

Minutes for 2024-05-29

Present
Gregg Kellogg, Ted Thibodeau Jr., Anatoly Scherbakov, Benjamin Young, Niklas Lindström, David I. Lehn
Chair(s)
Gregg Kellogg
Scribe(s)
Gregg Kellogg
Agenda
https://www.w3.org/events/meetings/08563546-30fd-497a-ac16-77f96e96e8c0/20240529T120000/
Topics
  1. Announcements and Introductions
  2. YAML-LD
  3. CBOR-LD
  4. JSON-LD Issue Discussion
  5. Open Discussion
Gregg Kellogg is scribing.

Topic: Announcements and Introductions

Ted Thibodeau Jr.: Previous meeting: https://www.w3.org/2024/05/15-json-ld-minutes.html
Gregg Kellogg: TPAC meetings mostly on Thursday 11:00-16:00
Benjamin Young: I'll send out TPAC scheduling information to the list.
Benjamin Young: I posted something to the charter repo.
https://github.com/json-ld/json-ld-wg-charter/issues/3 -> Issue 3 New normative document proposal: Supporting Multiple Languages in JSON (by BigBlueHat)
... There are some issues on the new charter, but I proposed some new normative documents.

Topic: YAML-LD

Anatoly Scherbakov: Working on test cases for processing YAML-LD docs with YAML-LD contexts.
... We talked about the JSON-LD API and how wit relates. I'll have something in a couple of days.

Topic: CBOR-LD

David I. Lehn: I haven't had time to push CBOR-LD issues forward.
Benjamin Young: If there are others here that have time to dig into issues, it would be helpful for people to ask questions.
... Questions help move the discussion forward.
Pierre-Antoine Champin: Interested, but short on time.

Topic: JSON-LD Issue Discussion

Subtopic: w3c/json-ld-syntax#425

https://github.com/w3c/json-ld-syntax/issues/425 -> Issue 425 how to "retype" rdf:JSON to geo:geoJSONLiteral? (by VladimirAlexiev)
David I. Lehn: Are we supposed to figure out how to fit this into our current spec?
... If we're not going to add it to the spec, we should create a best practice for how to deal with it.
Pierre-Antoine Champin: It looks like a niche problem; I wonder if there may be other such datatypes which expect JSON formatted strings.
... Clearly geoJSON literal is such a case, but are there others that would indicate the need for a general solution.
... A string works, but may be not as clean.
Gregg Kellogg: Similar to using custom XML as a subtype of XMLLiteral.
https://github.com/w3c/json-ld-syntax/issues/425 -> Issue 425 how to "retype" rdf:JSON to geo:geoJSONLiteral? (by VladimirAlexiev)
... My inclination is to not do something at this time.
David I. Lehn: It would be nice to be more user friendly.
... There should be some best practices for how to solve this, either as is or with hypothetical changes.
Benjamin Young: We may have mixed things when we modeled this. We wanted to treat a JSON object as an RDF literal, and maybe should have used something other than at-type
Benjamin Young: `@Type: rdf:JSON` and `@datatype: json`
... People can mint new datatypes, and if we don't have a place for people to dump this, we're limited.
Pierre-Antoine Champin: "Parsing mode" vs "RDF datatype"
... There are other things that are JSON objects that could have a different RDF type.
... We've sort of painted ourselves into a corner.
Niklas Lindström: Yes, it can be done, but when have we done too much?
... From a theoretical point of view it looks sensible, but from a pragmatic point of view, we need to see some other usages of it.
... If you have a property defined as geoJSON, you have a value which is an rdf:JSON datatype. When would that not be enough?
... Are there other similar datatypes?
Gregg Kellogg: Inferring a sub-datatype might be reasonable.

Subtopic: w3c/json-ld-api#513

https://github.com/w3c/json-ld-api/issues/513 -> Issue 513 Test compact#t0080 is wrong (by pchampin) [spec:bug] [ErratumRaised]
Pierre-Antoine Champin: It's been a while, so I'm not prepared to talk about this right now.

Topic: Open Discussion

Subtopic: Re-chartering

Pierre-Antoine Champin: I merged the PR I had created.
... When I had created, the intent was to re-charter the current group; not changing the dates, but the content. This was over-optimistic.
... We should revise the timeline and ask for a longer charter. That was not part of the previous changes.
... It occurred to me that we need to go through a new round of horizontal review.
... I'm not particularly worried about doing such a review.
... We should take this into account for the requested timeline of the charter.
... It will take more than a couple of weeks to re-charter.
https://github.com/json-ld/json-ld-wg-charter/issues/3 -> Issue 3 New normative document proposal: Supporting Multiple Languages in JSON (by BigBlueHat)
Benjamin Young: I was going to share the proposal for a new document.
... There's constant call for multi-lingual JSON; it's currently done poorly.
... JSON-LD is the one format that takes into account I18N considerations such as text direction.
Pierre-Antoine Champin: If we go there, all the more reason to go through horizontal reviews, the i18n group will want to have a say about this :)
... We should create a document that focuses on the value object and makes it into a first-class citizen that doesn't necessarily require other knowledge of JSON-LD.
... If you're implementing the spec in other areas you could reuse the pattern.
... I might want to filter for just english values, for example.
... It allows the use of this without bringing in the perceived overhead of JSON-LD.
Pierre-Antoine Champin: I like the idea. I think it's a shame to exclude language maps; they work nicely for this kind of thing.
... Perhaps adding the two options could be attractive to people.
... Should this be W3C vs IETF? Should it be in the JSON-LD WG?
... If this is published by the JSON-LD WG, don't we risk the counter-productive effect of people thinking they need the entire stack?
... Separating those things from JSON-LD is reasonable, but are we the right messenger?
Niklas Lindström: We talked about introducing other keywords for general use. We could consider a JSON-LD Lite.
... I would consider it a gateway to JSON-LD.
Gregg Kellogg: RDFa Lite did help, IMO.
... There's a danger of fragmenting our message.
Benjamin Young: I ended up using RDFa in contact data for an update to our webesite, and it tripped me up.
... I made it work, but it was non-obvious.
... The validators are the real tools people use to understand things.
... People usually don't want to model their data, but they want to do something anyway.
... Why do addresses need their own objects? Data modeling is hard.
... JSON-LD Lite could drift into encouraging the Cargo Cult school of markup.
... I'd rather see a JSON-LD 2.0 that lightens things than split the language.
... Language Maps require using some kind of header (context). The value object can exist without preprocessing.
... The JSON API spec could claim to support multi-languages by adopting this pattern.
... I didn't want people to be encouraged to use something else entirely, but want to move people towards better modeling.
... Language is a good story for JSON-LD. Language is a way to get people into JSON-LD.
... We should talk about language maps as a plug for full JSON-LD.
Niklas Lindström: +1 Makes sense (I could see other POVs, but it does)
... Webapp manifest does do language, but for the entire document.
Pierre-Antoine Champin: I agree with what bigbluehat said. I was thinking of encouraging design patterns.
... Anywhere these are used you have a multi-lingual string.
... I wouldn't add this to the charter without incubation.
... JSON-LD lite would be way to early to add to a charter.
Pierre-Antoine Champin: Note or REC, then?
Pierre-Antoine Champin: If we go forward with the value object deliverable, would be a note or a Rec?