- From: Gregg Kellogg <gregg@greggkellogg.net>
- Date: Mon, 8 Jan 2018 12:38:34 -0800
- To: Linked JSON <public-linked-json@w3.org>
Thanks to Dave Longley for scribing this week! The minutes for this week’s JSON-LD CG teleconference are now available: file:///Users/gregg/Projects/json-ld-minutes/2018-01-08/index.html. Full text of the discussion follows for W3C archival purposes. Gregg Kellogg gregg@greggkellogg.net JSON-LD Community Group Minutes for 2018-01-08 Agenda: https://lists.w3.org/Archives/Public/public-linked-json/2018Jan/0000.html Topics: 1. Introductions 2. Community Group details and process 3. Possible road to Recommendation? 4. Review current 1.1 work 5. JSON-LD Test Suite Resolutions: 1. same call every other week at this time. 2. accept issue 560 pending lanthaler approval Organizer: Gregg Kellogg and Rob Sanderson Scribe: Dave Longley Present: Gregg Kellogg, Niklas Lindström, Dave Longley, Robert Sanderson, Paul Warren, Eleanor Joslin, Paolo Ciccarese, Chris Webber, David I. Lehn, Herm Fisher, Paul Simmonds Audio: https://json-ld.github.io/minutes/2018-01-08/audio.ogg Gregg Kellogg: Voip 0302 is azaroth Niklas Lindström: Can’t hear you. [scribe assist by Gregg Kellogg] Niklas Lindström: Hm, no, I hear nothing... I'll try another device. Gregg Kellogg: Void: 0306 is Herm Dave Longley is scribing. Gregg Kellogg: How to scribe a meeting: https://www.w3.org/2009/CommonScribe/manual.html Robert Sanderson: We hear you, Niklas Niklas Lindström: Weird, I heard nothing; and lost the carriee. Paul Warren: Paul Warren also here on the same line as Herm Gregg Kellogg: Hi Paul! Niklas Lindström: Can you hear us now? [scribe assist by Robert Sanderson] Eleanor Joslin: Eleanor Joslin here on the same line as Herm and Paul Robert Sanderson: Great to have such a turn out :) Topic: Introductions Gregg Kellogg: I'm Gregg Kellogg, one of the editors on the 1.0 spec, and I've been doing a fair amount of work pushing the 1.1 spec forward. I'm mostly independent but also affiliated with Spec Ops. Robert Sanderson: I'm Rob Sanderson, I followed along the 1.0 work and we've become more deeply involved using JSON-LD 1.1 features in a couple of situations. Including IIIS, which uses several 1.1 features and new language map functionality and in the cultural heritage domain we use it in linked arts -- trying to provide linked open usable data with museum objects and so forth. Herm Fisher: I'm involved with XBRL a standard on XML base. Financial reporting for bank transfers, securities, for US/China/etc. Trying to move forward using JSON. Trying to get a better understanding of how to fit into the JSON-LD work. Paul Warren: Technical director at [fill] International, we're currently using JSON-LD in one of the drafts for XBRL. Eleanor Joslin: I'm a developer, here for the same reason as Herm and Paul. [missed]? David Newbury: Enterprise software architect with Rob, many of the projects we're doing involve JSON-LD in various forms. Paul Warren: Technical director at xbrl International David Lehn: I'm with Digital Bazaar, been working on JSON-LD for quite some time, been around for quite some time working on JSON-LD libs and helping with specs as I can. Dave Longley: I work with Digital Bazaar, I worked a lot on the 1.0 spec; I also maintain Javascript, Python and PHP JSON-LD libraries [scribe assist by Gregg Kellogg] Paolo Ciccarese: Focus on biomedical data, hence main interest in JSON-LD, worked with Rob. Chris Webber: I'm Chris Webber, I work with Spec Ops. Also a recently implementer of JSON-LD for Scheme, also did partial implementation for Guile. I may not been on many calls but around from time to time. Robert Sanderson: Proposal is that, at least for the mean time, we have a call every 2 weeks at this time. Trying to just get through the issues, get outstanding issues discussed, etc. Dave Longley: Sure, just que me in :) [scribe assist by Niklas Lindström] Niklas Lindström: Cue... Robert Sanderson: We'll probably keep using this call bridge, it has some nice tool integrations, but niklasl_ is having some difficulties so if others do as well we'll look into other options to make sure everyone can participate. Niklas Lindström: Yes PROPOSAL: same call every other week at this time. Gregg Kellogg: +1 Robert Sanderson: +1 Chris Webber: +1 Paolo Ciccarese: +1 David I. Lehn: +1 Herm Fisher: +1 Dave Longley: +1 Though i may not be able to join all of them :) Niklas Lindström: I’m Niklas Lindström, works for the National Library of Sweden And Niklas was also a major contributor to JSON-LD 1.0. RESOLUTION: same call every other week at this time. Eleanor Joslin: This time is workable for me, but slightly earlier in the day would be better Eleanor Joslin: OK Niklas Lindström: I was involved in JSON-LD 1.0. We use it all over the place in the new Library System at KB (National Library of Sweden) Gregg Kellogg: We may put out another doodle to see if we get a different response. Niklas Lindström: +0 Re time Robert Sanderson: We'll do the next call at this time just for consistency but there's a possibility for an earlier call. Topic: Community Group details and process Robert Sanderson: Mostly to emphasize that this is a W3C CG, it is not a Working Group. That means we can write specifications but they don't have any formal standing. We can have any process that we want, any of us can write a bunch of HTML and put it up somewhere, the most effective process to get JSON-LD 1.1 into standards track is to follow the process WGs use. Niklas Lindström: I will attempt to follow JSON-LD 1.1, as we’re dependent on 1.0 and want to see if 1.1 works for us. I have few cycles to spare at work alas, and a 1 year old daughter at home, so I’m pressed for call times as well. ;) Robert Sanderson: It gives the air of legitimacy and gives all the background and requirements a WG would need to take 1.1 forward. That means the use of github for tracking issues, the pull request model for making edits to documents and so forth. Robert Sanderson: That's how 1.0 was done so not new. Gregg Kellogg: https://github.com/json-ld/json-ld.org/issues Gregg Kellogg: Also important for CGs to operate as much as possible online to preserve the call time for more challenging issues. Gregg Kellogg: https://github.com/json-ld/json-ld.org/projects/2? Gregg Kellogg: The github repo where specs and issues are kept in IRC. Gregg Kellogg: Also a project tracker there. Gregg Kellogg: I encourage people to weigh in there, particularly if you support a PR or an issue, there is a nice thumbs up/thumbs down tool. Gregg Kellogg: We can try to drive things as much as possible that way -- but can also use email as well. Niklas Lindström: Just a quick re ”1.1”, will the CG version be called that? Gregg Kellogg: https://github.com/json-ld/json-ld.org/issues/548 Niklas Lindström: What I’m after is: will there be a ”JSON-LD CG 1.1” and *then* a possible WG 1.1? Robert Sanderson: We'll delay that question until we get to the issue for it. Any thoughts or questions about the CG details, etc.? Robert Sanderson: Anyone not been part of a CG or WG at W3C in the past? Paul Simmonds: I have not. Gregg Kellogg: Also an open issue, chartering a WG would be useful, but challenging. Eleanor Joslin: Me neither. Robert Sanderson: Thanks for participating! Gregg Kellogg: Right; but either way, is ”X.Y CG flavour” vs ”WG flavour” a risk? [scribe assist by Niklas Lindström] Robert Sanderson: There are a few oddities that you'll quickly understand how they work or why they happen. For the most parts W3C process is aimed at transparency and participating. Trying to ensure that everything is well documented. Niklas Lindström: (Thinking a little bit about WHATWG…) Gregg Kellogg: Niklasl, Open question. I have concerns about that to Robert Sanderson: For example, scribing everything everyone is saying. They can go back and look at the minutes later if not on the call. Robert Sanderson: This is more consistent and thorough than writing some bullet points later. Gregg Kellogg: OK, great! [scribe assist by Niklas Lindström] Robert Sanderson: This level of transparency is a relatively new. So the notion of having everything as public as possible is important. Gregg Kellogg: Relevant to that, these calls are being recorded. The practice has been, in the minutes, to include a link to the audio in the recording. Gregg Kellogg: So that's a fair warning. If there are specific things you don't want minuted please say so and we won't minute it. Gregg Kellogg: But it will still be in the audio call. Robert Sanderson: I don't imagine there would be much call for something so secret we need to pause the audio. Dave Longley: We can edit the audio later if super important but would prefer not to. Robert Sanderson: There is also a waiver that you need to sign to participate in the CG. If you haven't joined the CG, as an individual or representing your institution if your institution is a member, please do so. Robert Sanderson: The waiver is minimal, mostly don't sue. Robert Sanderson: Just some minor but important administrivia. Robert Sanderson: Any other comments on this topic? None Topic: Possible road to Recommendation? Robert Sanderson: If I recall correctly, JSON-LD 1.0 got to REC not by having a JSON-LD WG, but through the RDF WG. Gregg Kellogg: Correct. Robert Sanderson: That was some very fortunate timing because it meant that we didn't need to have a charter and all of the W3C members vote to create a WG with its own team contact and so on. Robert Sanderson: For 1.1, we are unlikely to be so lucky. Robert Sanderson: In terms of getting to REC, the initial barrier to entry is a bit higher. We'll need to get a WG up. Which means that, the more solid the work is that we do the more likely we'll get a WG. If it looks like a slam dunk at W3C, it's not rubber stamping of course, but if it's possible to show implementations, adoption, and good well-written specs out of the gate, then getting a WG approved is much easier. Robert Sanderson: At TPAC we had some brief discussions with team leads, Ralph Swick and Ivan Herman, and it's not only up to them, but they seemed reasonable confident with a path forward towards a WG. Robert Sanderson: The more work we do now, the more likely it is to happen in a timely fashion. In particular, the more members that are using it and building products around it, that require 1.1 features, that may be the most critical. Robert Sanderson: It's great to have the folks here from XBRL here as new adopters and the consistency and continuity of everyone else. Robert Sanderson: Gregg, others, want to talk about timelines? Gregg Kellogg: I guess there's a couple of ways we could continue. We could basically try to wrap up the work we're doing now in a bow so a WG could simply adopt, or we could push towards going to a WG earlier and defer. A WG charter would need to be created, it would be at least 6 months before a WG is chartered, if one is ever chartered at all. Gregg Kellogg: In these areas I think the CG output can be quite useful. It's not something that was available to people in the past, but CG output is more of a social standard, not the end of the world, but we'd like to see a WG. Gregg Kellogg: It would be nice to wrap things up with 1.1 and then work to set up a WG to ideally adopt that work or perhaps tag on to some other existing group to which the work is sympathetic. Robert Sanderson: I don't know how you thought about Annotation WG transition. Robert Sanderson: Paolo. Robert Sanderson: From my perspective, having a reasonably thorough spec -- there was space for additional work, if desired, seemed to be reasonably productive. Robert Sanderson: Paolo co-chaired the Annotation CG and then the WG, which concluded at the beginning of 2017. Robert Sanderson: The community group did not talk about protocol, there was no way of sharing annotations between systems, there was some good discussion in doing that in the WG context. That meant that the modeling side, while it was fully discussed, that was more about fine tuning than designing from scratch. Robert Sanderson: Having the design space for people to sink their space into rather than people thinking it was just a rubber stamping was helpful. Robert Sanderson: So the work we're doing here to get things started would be valuable. Paolo Ciccarese: I agree with what you said, I think what happened is -- we had some ideas before starting the WG. There were boundaries changed and pushed in some ways, but a good roadmap for the model and vocab to start. Not everyone was extremely happy about that. Paolo Ciccarese: Some people who didn't participate in the CG were offput because there were specs to work off of that had to be brought in. But to have something solid before starting is a good idea in general. Niklas Lindström: ;) Gregg Kellogg: A point of admin here, if you are in IRC and have a question or would like to contribute something, you can type "q+" to use the queue. Gregg Kellogg: You can do "q-" to take yourself off. Gregg Kellogg: Q? to see the queue. Robert Sanderson: I agree with Gregg that 6 months is the earliest to start on a charter and a target milestone would be TPAC in 2018. Robert Sanderson: Talking with potential staff contacts, etc. working through a charter to be proposed. Robert Sanderson: That would be November then, about 11 months, to propose a charter. Robert Sanderson: Having the support of W3C management is absolutely critical to getting things through. Talking in person is more productive than phone calls. Robert Sanderson: Any further thoughts or questions around the process? None Robert Sanderson: A former co-chair of mine said, if you're not, as chair, feeling like the silence is uncomfortable then you haven't waited long enough for people to ask questions :) Robert Sanderson: It is 20 to the hour, which gives us the right amount of time to review the 1.1 spec differences per Gregg's slides. Gregg Kellogg: TPAC slides https://json-ld.org/presentations/JSON-LD-Update-TPAC-2017/assets/player/KeynoteDHTMLPlayer.html#5 Topic: Review current 1.1 work Gregg Kellogg: I don't think we want to try and accomplish too much in the first week, doing an overview of what has transpired is a good place to conclude what we're doing here. Gregg Kellogg: I pasted in a link to the slide deck that describes the work that's been done since 1.0. Gregg Kellogg: There was quite a lot of time languishing between 1.0 and when we started getting more serious about moving forward. During that time there were a number of issues or errors, shortcomings of the 1.0 work. And new areas where people would like to see JSON-LD move in. Gregg Kellogg: That was enough to give us a general consensus of how to move forward. In particular, it was mostly about letting more JSON to take on JSON-LD meaning. And better ways to normalize the results. Additionally in the 1.0 time frame we were working principally, Dave Longley's work on framing, ... it has become quite important and more work as been done to improve the spec. Gregg Kellogg: We'd like to consider framing to be part of the 1.1 effort and part of the API instead of a separate spec. Gregg Kellogg: Looking at the slides, the versioning announcement creates a backwards compatibility effort. So some people are suggesting that we should do JSON-LD 2.0. Gregg Kellogg: Rather than "1.1". Gregg Kellogg: In this case we need to make sure that a 1.0 processor doesn't misinterpret the data -- and we have done that by introducing a version into the @context to cause a 1.0 to blow up rather than misinterpret the data. Gregg Kellogg: So the example does that -- in particular here to handle a @container that's different. Robert Sanderson: (Slide 7) Gregg Kellogg: In order to move more like JSON and use more objects and object keys to reference multiple things rather than arrays of things, we looked to increase the types of things for @container. Robert Sanderson: (Slide 8, identifier maps) Gregg Kellogg: Previously we could set @language, @index, on a term definition and that would cause a JSON-LD processor to interpret the object value of that term as a map. The keys would be language identifiers for @language and so on. When expanding it turns into an array of value objects with language tags within. Gregg Kellogg: We wanted to be able to do more things like that, for example, with ids and types. Gregg Kellogg: Examples on slide 8. Robert Sanderson: Eg the triple would be: <http://example.com/> schema:blogPost <http://example.com/posts/1/en> . Gregg Kellogg: Property nesting -- many uses of JSON include ways, where there's nominally an entity where properties aren't keys of the entity but part of some structure underneath (has no meaning in the data model). Slide 9 we see using that with "labels". Gregg Kellogg: This @nest feature folds values up a layer. Gregg Kellogg: When expanded, it's as if the nested keys are just properties of the top entity. Robert Sanderson: Hence the triple here is: <http://example.org/myresource> skos:prefLabel "This is the main label..." . Gregg Kellogg: Scoped contexts solve a number of issues. If you find yourself importing JSON-LD from another scope and you've been frustrated with having to add an embedded @context definition to define the @context in play for deeper values, this allows you to use a term definition instead. Gregg Kellogg: Slide 10 has examples. Gregg Kellogg: The "interest" term has its own context for its value without having to embed it. Gregg Kellogg: The feature is recursive so terms and is round-tripping. Particularly useful for framing. Gregg Kellogg: A scoped context in the frame definition will be applied to the framed results. Gregg Kellogg: You can also use scoped contexts on types. @Type scoped contexts are useful for "object oriented" like designs Robert Sanderson: (Slide 12, compact IRIs) Gregg Kellogg: Improving Compact IRIs is meant to make the use of curies less unexpected. Gregg Kellogg: There's a change where a term will only be considered to be used as a prefix if the prefix key is set or if the term ends in '/' or '#'. Gregg Kellogg: There are a variety of open issues after that. Gregg Kellogg: There are more open issues on github but they aren't tagged with 1.1. They may relate to the website or playground, but aren't critical towards updating the 1.1 specs themselves. Gregg Kellogg: Any question? Gregg Kellogg: Issues for us to consider soon -- do we want to stick with 1.1 and preserve 2.0 for a WG or stick with semantic versioning and move towards 2.0. Robert Sanderson: This discussion continued at TPAC in November. One thing that would be useful to inform it, would be what ... what are the actual breaking changes and breaking for whom? For example, a 1.1 processor that is using a 1.0 definition will work fine vs. a 1.0 processor that will break in some situation. Robert Sanderson: As was brought up in November, that's probably what we want in many of the situations. Robert Sanderson: Could you talk a bit more about what is broken by the changes? Gregg Kellogg: Probably not as much as you would want. One of the largest users of JSON-LD is Google. It is non-conforming though, they don't look at a @context and they presume it's the one that they've deployed. If you were to use this in there and expect schema.org partners to interpret it properly you'd be disappointed perhaps. Gregg Kellogg: Someone did report their processor breaking on a 1.1 version. But that's exactly the intent. But that raises the question about adoption. Herm Fisher: For XBRL we found a need to "certify" processors against a set of standard test suites. Gregg Kellogg: And how it would work. The side conversations I've had with Dan Brickley at Google -- they are not likely to dereference a @context would be ... it would be useful as a URI parameter. In the case when they are seeing a type definition in a script tag in HTML, if the type was "application/ld+json;version=2.0" that would give them the opportunity to make use of these features without actually having to dereference and process @contexts. Gregg Kellogg: In IIIF, Web of Things, things such as @id maps are required functionality, it might be slow adoption from some existing uses it could introduce cases where 1.0 compat isn't an issue. Herm Fisher: We've always had a set of standard test suites, that's part of every step of building changes to the spec. There was enough cowboy behavior that it was decided to provide a service that a product was part of the suite. Paul Simmonds: It's been nice to know how many conforming implementations have come out of the work as a result [missed?]. Topic: JSON-LD Test Suite Gregg Kellogg: https://json-ld.org/test-suite/ Robert Sanderson: We also have test suites for JSON-LD. From a process perspective, one of the W3C requirements is that there is a test suite that implementers can use to ensure that they have implemented the features in conformance with the spec. Robert Sanderson: It's a little fuzzy as to what a feature is and what tests are needed, but overall, it's very much part of the work that we need to do. Gregg Kellogg: There is a test suite that was begun in the 1.0 phase and has been continued. My practice has been, for every update to the spec, that there are tests to test for appropriate behavior. Gregg Kellogg: Those tests are marked with 1.1 so it's possible for a 1.0 processor to continue to run tests and not trip over things they aren't expecting. There is a quite comprehensive and maintained test suite. Gregg Kellogg: This test suite was also and the implementation reports were important for using a transition doc for showing wide implementation and at any time we choose we can put forward -- and do a call for implementations and accept implementation documents. Gregg Kellogg: That would allow us to show conformance we have. Gregg Kellogg: At this point the only fully conformant processor is my own, the Ruby processor, but work is underway on other processors as well. Robert Sanderson: It's the top of the hour. Any questions or comments? Gregg Kellogg: https://github.com/json-ld/json-ld.org/issues/560 Chris Webber: So I was originally pinged to join the call to suggest we take on the W3C's permissive software license for the standard. It didn't exist when 1.0 came out. I found I wanted to copy a lot of the language from the JSON-LD spec into the implementation I was writing and wanted to ensure it was legally feasible. PROPOSAL: accept issue 560 Chris Webber: I wanted to raise it possibly for next week since it was the reason I came onto this call. Dave Longley: +1 Chris Webber: +1 Gregg Kellogg: +1 Niklas Lindström: +1 Robert Sanderson: +1 Do you want the +1 here or on the issue? David I. Lehn: +1 David I. Lehn: Does this need approval from a wider audience? Perhaps from the other contributors? Robert Sanderson: It might be prudent to ping Markus about it. RESOLUTION: accept issue 560 pending lanthaler approval Robert Sanderson: Pending acceptance, as long as Markus is not -1. Robert Sanderson: Thanks everyone :) Robert Sanderson: So ... SIP on Mac ... a good free client? Gregg Kellogg: Linphone works well. Robert Sanderson: Okay, I think I have it set up. Will try next time.
Received on Monday, 8 January 2018 20:39:11 UTC