The W3C JSON-LD Community Group

Go Back

W3C Logo


Minutes for 2023-02-15

Topic: Announcements and Introductions

Anatoly Scherbakov is scribing.
Anatoly Scherbakov is scribing.
Gregg Kellogg: We had a WG meeting two weeks ago, one of items on agenda is to publish the YAML-LD spec.
Gregg Kellogg: One of items for this meeting is: can we make progress on moving towards a recommendation?
Benjamin Young: Need to finalize the draft so that the WG can pickup the work and finalize it as a recommendation.
Gregg Kellogg: Welcome Alexandra!
Alexandra Hitson: Hi! I work at Trying to learn more about the industry, especially about JSON-LD. I am from North Carolina, US.

Topic: YAML-LD

Gregg Kellogg: When we last worked on YAML-LD spec we split the Use Cases document. I have been busy in WG lately. Looking for people to take up the work on YAML-LD.
Gregg Kellogg: Primary thing WG is interested in is publication of the core spec. What can we do to get that into shape?
Gregg Kellogg: Spec is pretty detailed, it supports both basic & extended profiles. I feel extended profile is too ambitious to be specified right now.
Gregg Kellogg: Tags and IRIs with hashes in them are not widely supported in libraries and tooling
Gregg Kellogg: The extended internal representation and parsing of YAML documents including special tags should probably be moved into an informative appendix. Not be a part of core feature set.
Gregg Kellogg: We should be able to just leverage JSON-LD core processing methods for YAML-LD.
Gregg Kellogg: We need people interested in bringing the document into a final state when, as a CG, we can publish it as a final report.
Gregg Kellogg is scribing.
Anatoly Scherbakov: For some time, I didn't understand the benefits of tags and the features of the extended representation.
... That was because I was mostly relying on contexts, but at some point
... I got an impression that using such tags would be beneficial for improving an editor's experience.
... Since the spec postulates the existence of an extended internal representation,
... we would need special processing rules for extending an extended document to the plain document.
... That might replace all objects that are represented as tags as normal JSON-LD representations.
... Those questions could be mute if we decide to postpone until later.
Gregg Kellogg: I looked into transforming Extended format straight to JSON-LD with value objects. How do we get back to original YAML representation with tags though?
Gregg Kellogg: Idea of the Extended representation was that we'd be able to represent literals with tags. On JSON-LD side we can process them as RDF literals but we can get them back to tags for YAML.
Gregg Kellogg: Conversion to JSON-LD value objects might be an extension to a library or a CLI. My implementation is basically a subclass, a JSON-LD Extended Loader.
Gregg Kellogg: It provides DocumentLoader for YAML-LD and extensions for core methods.
Gregg Kellogg: That is a route that can be taken. These issues are open enough and we could continue on that forever. We might instead prefer to leave them into an not normative sections and invite people to experiment.
Gregg Kellogg: We need people to continue work on this spec because I am busy with other work and it is unclear how to proceed unless people step up.
Anatoly Scherbakov: I tried to do some contributions, and I would have said I'm eager to do that, but I'm concerned about my availability to contribute.
... Since LD is a spare-time effort for me, and I have my own project that touches YAML-LD, but it also consumes a majority of my time.
... I could try doing something, but I can't promise.
... Maybe we can do some small tickets to split the work into small pieces, such as "remove extended profile into an informative appendix".
Ted Thibodeau Jr.: I will continue polishing via suggestions on PRs, and may be able to take a whack at the doc, section-by-section, but cannot make promises of delivery date, as things stand [scribe assist by Ted Thibodeau Jr.]
Gregg Kellogg: Creating small issues is probably a good idea. It can help engaging people and encourage to participate
Gregg Kellogg: Question is who is going to actually drive the effort. There is no deadline, WG is recently re-chartered, it has two years. There might be updates to core JSON-LD spec during that time
Gregg Kellogg: Work is also going on in RDF-star.
Benjamin Young: Maybe many of us are here not only because of YAML-LD. I wonder if key things people are interested in are in YAML-LD or perhaps in JSON-LD, RDF-star or other topics.
Gregg Kellogg: A lot of work is going on around JSON-LD, there are outstanding feature requests; JSON-LD-Star is being worked on but blocked on RDF-star
Gregg Kellogg: Verifiable Credentials group uses CBOR-LD, that's only an ongoing effort. They also want to arrive at some form of a standard.
Gregg Kellogg: What brings people today? What are we interested in working on?
Gregg Kellogg: Ted?
Ted Thibodeau Jr.: My fingers are on almost anything Identity or LD related. We use RDF and LOD in my company.
Ted Thibodeau Jr.: I do everything except code.
Juuso Autiosalo: I still want to see YAML-LD published as a recommendation. For me, Extended representation isn't needed much though :) Unfortunately I am overloaded with work currently
Juuso Autiosalo: I would like to find time to work on this spec but I am not familiar with the internal representation and the tooling around the spec. In my university, I could try asking if someone else might be interested in contributing.
Gregg Kellogg: Contributing to issues and PRs, reading changes, asking questions, making suggestions is a great way to contribute. Basic internal representation is basically just JSON.
Gregg Kellogg: We can do a lot with just that. If you have students they might be interested in test contributions, implementations — all suitable projects for students.
Gregg Kellogg: This is probably not enough for a doctoral thesis but still might be interesting.
Juuso Autiosalo: Have we already decided we do not do the extended representation or is this a decision to be made still?
Gregg Kellogg: There is not enough people here to decide finally. We can create an issue and invite everyone to comment on that.
Gregg Kellogg: I feel trying to accomplish less leads to higher probability for success.
Gregg Kellogg: Extended representation might be something the WG might want to work on.
Ted Thibodeau Jr.: Small bites are almost always done faster, better, more, than big bites [scribe assist by Ted Thibodeau Jr.]
Alexandra Hitson: Trying to understand the basics and see what further topics I might be interested in. Listen, see what people are talking about and get acquainted with what's happening. Still interested to contribute.
Gregg Kellogg: That's how many people take part in such activity, and all contributions are appreciated. Rarely it happens that I make a PR and there is no feedback to it.

Topic: Next Steps

Benjamin Young: Thank you all for being here. There is a lot of JSON-LD out there, great when people get interested in the technology. I'd like Mastodon folks be here too.
Gregg Kellogg: JSON vs JSON-LD people seems to be permanent thread at Verifiable Credentials
Ted Thibodeau Jr.: +1 Recruit via "AMA about JSON-LD, YAML-LD, etc."
Benjamin Young: JSON-LD is not a finalized tech. It still could benefit from dev evangelism and community outreach. CG are the closest thing to this. I've been wondering about changing the model of such meetings to invite more people, even new people, those who're maybe not ready to contribute to the spec right away but are interested in the technology.
Benjamin Young: Get more people in the group and more people to do the work.
Gregg Kellogg: This is a great idea. Mastodon is exposing people to JSON-LD but kind of indirectly. Encouraging people to participate would be great and drive some momentum.
Gregg Kellogg: Next baby steps might be to make a couple little changes, start the pebble rolling down the hill. We can meet, say, in 4 weeks time — 15 of March
Gregg Kellogg: I will update the calendar and post the agenda soon to boost visibility
Ted Thibodeau Jr.: I expect to be here 2023-03-15 [scribe assist by Ted Thibodeau Jr.]
Juuso Autiosalo: Thanks Benjamin for welcoming us to this community. I talked about JSON-LD inside our organization and advertisiing around our reearch group. It seems to be a great fit for what we're doing. Thanks for all the work you've already done.
Ted Thibodeau Jr.: There was a discussion about structured data islands into web pages. We have a thing named Structure Data Injector which puts into web pages chunks of structured data, and are then digested by search engines.
Gregg Kellogg: Embeds in HTML pages are probably the widest deployment of JSON-LD
Benjamin Young: YAML is not a great thing to stick into HTML because HTML editors don't like whitespace.
Gregg Kellogg: JSON-LD in <script> tags wasn't a thing first when JSON-LD started development. Not before adoption.
Ted Thibodeau Jr.: Big challenge is that you have escape things differently when embedding into <script> tag, that happens for Turtle and JSON-LD both.
Ted Thibodeau Jr.: We have to put into <script> and comment it. It is still being parsed.
Benjamin Young: HTML spec has Data blocks. If there is a <script> tag which doesn't specify it is JS then it is going to be ignored.
Benjamin Young: Data-block-extract
Benjamin Young: It might depend on whether it is XHTML vs HTML etc, but it's an existing feature that has been there for decades. I built a tool to extract those data blocks from CLI
Gregg Kellogg: CDATA isn't really a part of HTML anymore, and XHTML is considered dead. But the idea that you can put stuff into <script> tag and it doesn't require HTML escaping is promising.
Ted Thibodeau Jr.: Whitespace remains in the source. You can see all the whitespace in the source. Viewing it as HTML will of course eliminate all the whitespace.
Benjamin Young: YAML block might be indented, shifted over together with the <script> tag — and that might confuse the YAML parser.
Gregg Kellogg: YAML has to start on a new line, not immediately after the <script> tag.
Benjamin Young: HTML has to strip whitespace. JS can live with that.
Benjamin Young: Code editor auto indents HTML. If I type YAML into it then I have to fight the editor
Gregg Kellogg: YAML processors should work off the first line they see. It can't be completely left justified.
Ted Thibodeau Jr.: I had once to split out such leading characters.
Gregg Kellogg: It still should work if all the lines have the same indentation level
Gregg Kellogg: If you insert Markdown into HTML then you have the same indentation considerations
Ted Thibodeau Jr.: Damn you technology!!!
Gregg Kellogg: Thanks everyone for showing up! get comments and PRs. we will meet again in 4 weeks.