The W3C JSON-LD Community Group

Go Back

W3C Logo


Minutes for 2023-11-15

Gregg Kellogg is scribing.

Topic: Announcements and Introductions

Topic: YAML-LD

Anatoly Scherbakov: No YAML-LD updates this meeting.
... We need to advertise the spec to a wider audience, and do a self-review.
Gregg Kellogg: I'll send an update to the semweeb mailing list.
Benjamin Young: I'd like to see about adding CBOR-LD to the CG with the same intention
Pierre-Antoine Champin: +1 To that

Topic: JSON-LD Issue Discussion

Subtopic: -> Issue 421 JSON-LD Context similarity to XML External Entity attack (by hmottestad)
Gregg Kellogg: Something for BP, which needs to be published.
Benjamin Young: Latest Best Practices Note

Topic: Spinning up Working/Maintanence Group

Pierre-Antoine Champin: We don't need to make a decision on a call, it can be done on email.
... Not clear what quorum is required. But, we probably need a +1 from some number of the members group
Ted Thibodeau Jr.: I think the CG and WG are effectively the same group, but the agendas should be separate.
Anatoly Scherbakov: Don't think I am a member of the WG :)
... CBOR-LD is not in a charter, but a CG could agree to add it to its plate.
... Then it needs to get past the WG and needs to be added to the charter.
Benjamin Young: In the other delieverables, it mentions YAML, but that thay're non-normative.
... By and large, as it's actually implanted, and the CG members cross over with the WG.
... We need to refresh our charter periodically, in any case.
... In the list of Other Delivereables, YAML-LD, CBOR-LD, and the best practices note.
Pierre-Antoine Champin: Technically, this meeting appears in the calendar of both groups. I'm not sure if that's enough process wise to do work here, and the agenda hasn't been explicit.
... The WG charter says we can publish non-normative docs, but I had a discussion about the use of notes by some groups to bypass rec-track documents.
... Therefore, I think the YAML-LD spec is intended to be REC-track, not a note.
... We could publish a note with some guidance, but a proper spec needs a re-charter.
Benjamin Young: I'm not trying to be sneaky. The CG stuff caries similar weight to a WG note.
... But, I agree that both of these should be REC-track notes.
Pierre-Antoine Champin is scribing.
Gregg Kellogg: I thing we should have an *explicit* WG meeting, in the normal time slot,
... with an agenda that explictly looks towards the anticipated final report for YAML-LD and long standing work on CBOR-LD,
... as items to take up in a new charter.
Ted Thibodeau Jr.: +1 For having such explicit WG (and CG) agenda items whenever possible
... We could probably address that in a 2 year new charter.
... For maintenance of the core spec, we are pending progress in the RDF-star WG.
... If we can organize a WG discussion in 2 weeks, that includes the CBOR-LD note,
... we would need someone who would be up for it.
Benjamin Young: Actually, dlongley asked me to put CBOR-LD on the agenda.
... To get to REC, we'd need another implementation.

Topic: CBOR-LD

Gregg Kellogg: As I recall, the CBOR-LD spec by Digital Bazaar is not entirely congruent to YAML-LD.
... It does not describe how to map CBOR into the internal representation, etc.
... It focuses on how to build a context that makes the data more compact in CBOR.
Pierre-Antoine Champin: +1 To that, we have two specs; I contributed to one, and the DB version should be another spec.
... We should have both a systematic conversion, and something on the context.
David I. Lehn: I agree with what's being said. DBs use of CBOR has been mostly for compression.
Pierre-Antoine Champin: Both specs are and
Ted Thibodeau Jr.: "JSON-LD 1.1 in CBOR"--
Ted Thibodeau Jr.: "CBOR-LD 1.0"--
Gregg Kellogg: Do we want to have this discussion in the CG first? Or do we call a meeting for rechartering right now? [scribe assist by Pierre-Antoine Champin]
Pierre-Antoine Champin is scribing.
Anatoly Scherbakov: Not sure how to organize the process, but I'd welcome a CBOR-LD standard.
... I don't have experience with CBOR. Is there some intention to include it in PyLd.
David I. Lehn: I think it would be separate.
Gregg Kellogg: A Python library for YAML-LD would rely on PyLD, but keep the YAML-specifics out of it
... That's how I did it in Ruby.
... Requires some work to change the content negotiation on remote context retrieval.
Niklas Lindström: .. LDProcessor<WireFormat> ;)
Benjamin Young: I'll organize the next meeting as a WG meeting.

Topic: JSON-LD issues

Subtopic: -> Issue 421 JSON-LD Context similarity to XML External Entity attack (by hmottestad)
Gregg Kellogg: Any thoughts on this?
... My thought was to make it an item for BP.
David I. Lehn: Are there examples of the problems? It's kind of generic
... I asked for specific attacks, but it's hard to give advice when you don't have specifics.
... There are specific attacks that might not apply to us.

Subtopic: -> Issue 420 feature request: support `$schema` as an alias for `@context` (by pchampin)
Pierre-Antoine Champin: I explained the rationale in the issue.
... my idea is that usually when you want to specify a JSON as a schema, it would be good to provide both a schema and a context.
... Both formats are nice enough that they could work together. The context is ignored by schema, and the schema is ignored by the context.
... We could say that instead of providing `@context`, you could provide `$schema`, which would have a similar affect.
... There are sometimes confusion about contexts vs schemas, that this might help with.
... Putting them the same file, though, could be confusing.
Gregg Kellogg: You can also put a context and a JSON-schema in the same file. Maybe you could put the schema also in the same file!
... I had similar discussion with Mastodon people, about how they are annoyed by the JSON-LD parts.
Niklas Lindström: +0.9 For schema referencing context, -0.9 for context referencing schema... I think. :)
... This adresses issues that have long been raised in the community.
Anatoly Scherbakov: My feeling is that this change would increase the confusion by mixing concerns.
... Validation and expansion are separate concenrs.
... I think the schema can be created from a context, but not the other way around. Combining them may be counterproductive.
Benjamin Young: I'm a -1 on this issue, as affordances, they imply different formats and intentions.
... We'd also be squatting on another namespace.
Niklas Lindström: ... RDF Schema :P
... JSON schema is just surface syntax enforcing.
... The Context is only about term mapping
... Things like framing get to the point where they provide surface syntax, but people have long wanted to use a context as a schema.
... Now with `@protected`, it looks even more like a schema.
... They get close, but they're not the same thing.
... I think there needs to be more work to describe surface syntaxes.
... It ends up, you need both. There are other tools that have a way to express both, and then generate a context and a JSON schema, along with RDF/OWL.
... This feels a bit like transformation conversations from before.
... There are people in the JSON schema community that would like to address this.
... Papering over it is scary; unless a JSON-LD 2.0 makes substantial changes.
... Weily used to have a JSON-LD context file where we baked in everything.
Niklas Lindström: I agree with what bigbluehat said. I'm not sure we'd every be able to do such a think in JSON-LD 2.0.
... I'd like to keep the issue open, because it tracks the needs of the larger JSON community.
... For now, I'd rather see something to clarify what this means.
Benjamin Young: A spec that uses both together quite happily:
... I think of Linked Data as drawing maps, vs XML laying railroad tracks.
... I'd like to keep this open for now, and try to get developers more into the RDF mindset.
Pierre-Antoine Champin: I agree with the concerns that are raised, and that it is conflating concerns.
... However, the two things aren't completely decoupled, and the uber-formats that exist are an indication of that.
... Maybe SHACL shapes have a place.
... It can be dangerous, but has some nice features.
Benjamin Young: I dropped a link to the JSON Schema in RDF for Web of Things.
... Every thing description is both a schema and a context.
... If you wanted to take a context with `@protected` and through in some schema, it would fail right now, but it could be interesting to explore.