Gregg Kellogg: There have been a number of emails going by in the last day or so about publication of JSON-LD. Invited RDF WG chairs and W3C staff members.

Gregg Kellogg: As a result of those conversations, we have a proposal to take JSON-LD to REC right after RDF 1.1 specs go to PR.

Gregg Kellogg: This is mostly dealt with by the staff. We will want to get some resolutions down to smooth the process.

Gregg Kellogg: The only slight objection to this path was from Ivan. He felt that there would be more "thunder" if JSON-LD would come out with the RDF 1.1 documents. Both Guus and David agreed that JSON-LD is something different, and there are advantages to having it out on its own separate form.

Gregg Kellogg: Rather than trying to do this at the same time, they said they'd support doing the PR and then REC for JSON-LD in January.

Gregg Kellogg: Is there any anecdotal evidence to take JSON-LD to REC earlier than RDF 1.1 stuff?

Manu Sporny: I've been having some offline emails with several large deployers of JSON-LD [scribe assist by Gregg Kellogg]

Gregg Kellogg: ... (private conversations) asking about status so that they can signal inside about going to production with several tools and projects

Gregg Kellogg: ... They don't want to be in a position of releasing a product which would then need a change.

Gregg Kellogg: ... This is why I'd like to get it published sooner. There's no new work, even if it doesn't go to REC, it doesn't affect anything, but it has a practical impact on people that are using the technology.

Gregg Kellogg: ... The normative references to RDF docs are fairly academic, and wouldn't affect an actual web developer.

Gregg Kellogg: ... Also, if RDF were to change drastically, there would be a bigger issue in removing things rather than keeping them in.

Gregg Kellogg: ... Delaying publication of JSON-LD seems more academic rather than preventing something that could actually happening.

Gregg Kellogg: ... I think there's general agreement to take JSON-LD to REC mimediately after RDF Concepts/MT publication

Gregg Kellogg: +1

Dave Longley: +1

Manu Sporny: +1

Paul Kuykendall: +1

Markus Lanthaler: +1

David I. Lehn: +1

Gregg Kellogg: It'll be good to get the work out there. Lots of hard work.

Gregg Kellogg: A couple of work items have been left hanging due to the drive of getting the core work out.

Gregg Kellogg: JSON-LD Framing has come up lately. RDF Graph Normalization is also important. A streaming API has been discussed as being useful for very large dataset dumps where in-memory processing requirements are prohibitive.

Gregg Kellogg: There has been talk about something like the indexing mechanism, but a way to ignore a layer of keys.

Dave Longley: We have a number of lingering issues on github, we have a few more ideas that we could add to the @context.

Dave Longley: We wrote these down on github as issues before.

Paul Kuykendall: Has there been any other discussions w/ Microsoft on OData alignment.

Gregg Kellogg: No

Markus Lanthaler: Nope.

Markus Lanthaler: He said he'd send OData examples to the list and we'd show how it could be modeled in JSON-LD.

Markus Lanthaler: We didn't get that mail, unfortunately.

Gregg Kellogg: There has been some discussion about making JSON-LD compatible w/ HAL. Another place this came up is in the Microdata definition - we could try to align JSON-LD w/ Microdata.

Gregg Kellogg: I think we're seeing people wanting to use keys and subobjects to compartmentalize data.

Markus Lanthaler: Yeah, it doesn't work well if you have properties that don't mean anything but you want to still have a connection between two objects.

Gregg Kellogg: What is the kind of framework underwhich we can do extensions to JSON-LD w/o breaking the existing spec.

Manu Sporny: Two main types of extensions: algorithmic, which don't break the spec too much. Syntax changes have a chance to destabilize. If we start making syntax additions we'd be needing a JSON-LD 1.1 or 2.0. These should be discussions around extensions, and not push things until we're sure we want to make such changes. Framing is something we can play around with in the API, same thing with Normalization and Streaming. Talking about compartmentalization of object linking, we inevitably get into talking about syntax. [scribe assist by Gregg Kellogg]

Manu Sporny: There are two ways of approaching this compartmentalization feature.

Gregg Kellogg: Are JSON-LD Frames a part of the syntax?

Markus Lanthaler: They're different... different media type, etc.

Gregg Kellogg: Hmm, yes, but we do have code in the main algorithms that deals w/ frame data, right?

Markus Lanthaler: Yeah, but we don't need to stick to that mechanism.

Gregg Kellogg: We do have a way of specifying which processing rules to use.

Markus Lanthaler: But not within the document.

Markus Lanthaler: A lot of those mechanisms come from converting other data to JSON-LD. We don't need to pollute the syntax w/ conversion of old documents to JSON-LD.

Markus Lanthaler: I think many people just want to use HAL is because of the richer toolset.

Markus Lanthaler: I think we should concentrate more on tools/libraries around JSON-LD.

Manu Sporny: We should work on tooling to solve issues rather than looking at the standards right now. Make it easy for web developers. [scribe assist by Gregg Kellogg]

Paul Kuykendall: I agree. At what point are we betting on winners and losers re: mindshare.

Paul Kuykendall: The tooling is a better option, make it as agnostic as possible.

Dave Longley: I think we can move more quickly by working on tools instead of working on syntax.

Gregg Kellogg: One possibility would be to look at JSON-LD Macros as the basis for some common tooling. It allows for ad-hoc modification to get into a JSON-LD format. Anyone spent any quality time w/ his work?

Markus Lanthaler: Not really. Just syntactic transformation is the least difficult part. The whole idea of Linked Data is much more complicated than transforming JSON representations. You need tools to show you the power of using such an approach.

Markus Lanthaler: You need to build on top of it. Something that looks simple, but has big improvements over HAL or something like that.

Paul Kuykendall: If you want to put extensions, where do you do that? Should we say how to do that? We can keep the extensions more isolated/sandboxed.

Manu Sporny: We don't want to restrict people by telling people how to do extensions.

Paul Kuykendall: That's not what I mean, how do we tell people they can work with the community. Which thread do you pull on first? We need docs to tell people how to extend JSON-LD.

Paul Kuykendall: We can maybe use HAL as the example, to show some of the power of JSON-LD.

Manu Sporny: Agreed.

Gregg Kellogg: yeah, it could be a blog post. Someone could do that and then we could create a CG note from it.

Markus Lanthaler: The problem with those documents is that nobody wants to write it. We can reach out. Answer people where they're asking questions.

Markus Lanthaler: I don't understand why they can't publish a JSON-LD Context.

Manu Sporny: I don't know if this is something they're interested in fixing.

Gregg Kellogg: Dan Brickley was concerned that publishing a JSON-LD Context would create many millions of requests to Google systems.

Gregg Kellogg: We may have to build schema.org into our processors.

Markus Lanthaler: Is the Context ready?

Gregg Kellogg: He needs a Python program to translate schema.org vocab to JSON-LD Context. He needs something like that.

Markus Lanthaler: I think niklas already hacked something together for that?

Gregg Kellogg: I'll send an email out to public-vocabs.

Niklas Lindström: .. https://gist.github.com/niklasl/7873635

Dave Longley: We could put the context in ./well-known/context.jsonld

Gregg Kellogg: q?

Niklas Lindström: I sent this to schema.org and Dan, didn't receive any other response.

Niklas Lindström: Yes, I think it's mostly there. I think Dan wanted to discuss what we need wrt. coercing things that are strings vs. things.

Niklas Lindström: I adapted the script so it behaved in a sane way by default. Don't know if it's enough.

Gregg Kellogg: He may want commitment on tool providers to do what they can in order to limit excessive invocations of that URL via caching.

Gregg Kellogg: Maybe you guys can follow that and emphasize it so that we can get a commitment to get it done.

Paul Kuykendall: We need to list JSON-LD tools in a prominent place.

Gregg Kellogg: I'm not sure where else we might publish that. Maybe on the JSON-LD landing page.

Paul Kuykendall: The mailing list can be byzantine.

Gregg Kellogg: Can we access the stats for the site?

Manu Sporny: The whole reason DB started with JSON-LD is because of web payments. And, of course, because of people on this call it took on a life of it's own. Now that 1.0 is out there we need to apply it to work. Framing's not that useful, but normalization is quite important. We'll focus more on normalization; framing will be done eventually, but don't have much time to work on it. I don't think we'll have much time to work on tooling otherwise. From our perspective, we're willing to put time in to RDF Graph Normalization primarily, and bring that to REC. After that, framing and other tooling. [scribe assist by Gregg Kellogg]

Gregg Kellogg: ... I imagine that there is other tooling as a by-product of web payments (validation, etc.) and as a side-effect we'll create some tools to verify hashs, signatures and so forth. The action we'll take is where to take Normalization next and getting it specced out. There are web site updates, HAL conformance, and schema.org relationships.

Manu Sporny: I could try to put aside a day to work on the json-ld.org site. I thought about reaching out to the local university and offering to do a mentorship to get credit and work on JSON-LD stuff. We could potentially get 4-5 people to contribute. Also GSoC coming up again next summer. [scribe assist by Gregg Kellogg]

Gregg Kellogg: If we could start a conversation on tooling, we can see what updates to the website need to be done.

Gregg Kellogg: Next meeting - next Tuesday is off, after that is New years. Let's meet again on the 7th of January.

Gregg Kellogg: Let's see if we can get some progress by that point.

Gregg Kellogg: We should have regular but infrequent meetings to keep pressure on JSON-LD to keep moving. We need to get more people involved, to share the burden and for this to take off, it can't remain the original core contributors. It needs a life of its own.

Markus Lanthaler: pkuyken, can you hang on a second after the call.. have a quick question

Gregg Kellogg: I'll send out a meeting Agenda for the next meeting before the 7th of Jan.