JSON-LD Working Group Telco — Minutes

Date: 2020-03-20

See also the Agenda and the IRC Log

Attendees

Present: Gregg Kellogg, Rob Sanderson, Ruben Taelman, Adam Soroka, Pierre-Antoine Champin, Benjamin Young, Ivan Herman, Tim Cole, Harold Solbrig, David I. Lehn

Regrets:

Guests:

Chair: Rob Sanderson

Scribe(s): Pierre-Antoine Champin, Rob Sanderson

Content:

1. Announcements

Rob Sanderson: any update on editorial changes?

Gregg Kellogg: some non-normative changes
… @prefix=false millions of need some precision in the syntax document
… requires some examples
… Most other pull-requests are about tests.

Pierre-Antoine Champin: One more: clarification in the “inverse context creation” algo;

Gregg Kellogg: One more: clarification in the “inverse context creation” algo;

1.1. no-response API editorial changes

Gregg Kellogg: some steps were reordered, and an “otherwise” was missing.

Rob Sanderson: we resolved to close a number of issues if we did’nt hear about them until today,
… and we didn’t.
… Greg (kasei) did some changes related to the issues he submitted
… which fixed some tests, but broke others.
… He was happy with those issues being closed.
… Greg W was worried that nobody implemented Framing from scratch.

Gregg Kellogg: Framing has changed quite a lot since 1.0.

1.2. Other announcements / reminders?

Gregg Kellogg: there are signs of life of various repositories.

Harold Solbrig: Some work is happening on RDFlib.

Ivan Herman: I recently hit some issue with RDFlib,
… it seemed that nobody was touching the core of the lib.
… It does not work with recent versions of Python 3.

Harold Solbrig: yes, there has been much activity recently.

Pierre-Antoine Champin: One of my old PRs for rdflib was accepted :)

Rob Sanderson: any other implementation?

Gregg Kellogg: I switched from jsonld-js to pyld,
… solved a few issues there.

Rob Sanderson: another issue we ran into.
… we have a pipeline generating many JSON-LD files,
… converting them to RDF and storing them in a triple store.
… Context processing is taking a lot of time.
… Some caching would be beneficial.

Gregg Kellogg: we have that guidance stating that implementations should cache contexts,
… something that was asked in particular by danbri about the schema.org context.
… Standard HTTP caching can be used,
… but my implementation also has an ad-hoc cache, storing contexts in their internal representations.
… We might recommend a set of techniques for doing that.

Harold Solbrig: we find ourselves modularizing contexts a lot,
… rather than having one big context covering everything.
… So we rely on caching a lot.

Rob Sanderson: gkellogg, Greg W and I will have a call about this next week.
… If anyone wants to join…

David I. Lehn: jsonld.js has some support for caching too

Gregg Kellogg: in jsonld.js and pyld, the internal representation of contexts is basically a dictionnary.
… caching the result of “context processing” is quite easy.
… last week, I said I wanted to merge the expand and toRdf test suites.
… there has been a big PR unifying the naming, which should make this easier.

2. Technical Issues

Rob Sanderson: I don’t think we have any tech issue?

Harold Solbrig: I found one yesterday, actually…

Rob Sanderson: https://github.com/w3c/json-ld-syntax/issues/339

Harold Solbrig: the term IRI ends with a gendelim, so it looks like a valid prefix
… RDFlib uses it as a prefix, even when @prefix=false is set.

Gregg Kellogg: it is an internal representation issue, it has nothing to do with JSON-LD
… the API document is clear, @prefix=false prevents the term to be used as a prefix
… in expansion or compaction.
… But the syntax documents needs some clarification.

playground: http://tinyurl.com/rpddoss

Harold Solbrig: I don’t go to the API doc for this kind of issue.
… Fixing the syntax document is good for me.

Rob Sanderson: this kind of example can be useful for having human-readable types
… Is it worth puting an editorial note about spaces?

Gregg Kellogg: we made a provision for the empty string, and terms ending with a colon,
… but where would we stop? What about tabs?

Rob Sanderson: acj hsolbrig

Harold Solbrig: this kind of example will creep up more and more.
… When importing data from spreadsheets, spaces (in headers) may be common.

Gregg Kellogg: we could put some restrictions on terms used as prefix…
… In CSVW we addressed this kind of issues, using URI patterns.

3. Notes

Rob Sanderson: rubensworks, any progress on the Streaming note?

3.1. Streaming Note

Ruben Taelman: I opened a new PR this morning. Not sure it needs to be reviewed.
… Once merged, I will ask people to go to through the whole doc.

Ruben Taelman: https://github.com/w3c/json-ld-streaming/pull/4

Ruben Taelman: https://w3c.github.io/json-ld-streaming/

Ruben Taelman: Everything I want to have in that note is present now.

Pierre-Antoine Champin: .. Feel free to open an issue if you think something is missing.

Gregg Kellogg: have you considered having a kind of test suite for streaming?

Ruben Taelman: we could add a few tests (@context and @type not in the correct order)
… Let me add it to my list.
… That suite would be quite small. There are only a few restrictions.

3.2. CBOR

Pierre-Antoine Champin: no activity there on CBOR, France is stuck in confinement right now

3.3. Best Practices?

Rob Sanderson: [crickets]

Harold Solbrig: we would be interested in getting access to the original order in JSON lists,
… in order to apply our own sorting algorithms.
… But I understand this goes beyond the scope.

Benjamin Young: https://www.w3.org/TR/csv2json/

Gregg Kellogg: https://www.w3.org/TR/tabular-metadata/

Gregg Kellogg: you should look as CSVW, it might help address your use case

4. AoB?

Ruben Taelman: https://www.rubensworks.net/blog/2019/03/13/streaming-rdf-parsers/

Ruben Taelman: a while ago, I wrote this blog post.
… Would it be valuable to link to that within the note. Is it appropriate?

Ivan Herman: It is a note, so I don’t think this would be a problem.

Ruben Taelman: I will include it then.

4.1. Streaming, redux

Ruben Taelman: Any specific process for finalizing the note?

Ivan Herman: we have to have a formal resolution, then go through the webmaster.
… This is easy, mostly administrative (getting an ID).

Rob Sanderson: any preferred short name for it?

Gregg Kellogg: json-ld11-streaming?

Rob Sanderson: +1

Ivan Herman: +1

Ruben Taelman: +1

Proposed resolution: Short name for streaming note should be json-ld11-streaming (Rob Sanderson)

Gregg Kellogg: +1

Rob Sanderson: +1

Adam Soroka: +1

David I. Lehn: +1

Tim Cole: +1

Pierre-Antoine Champin: +1

Ruben Taelman: +1

Harold Solbrig: +1

Benjamin Young: +1

Resolution #1: Short name for streaming note should be json-ld11-streaming

Rob Sanderson: anything else?

David I. Lehn: here’s some JSON-LD helps save the world news: http://blog.schema.org/2020/03/schema-for-coronavirus-special.html

Benjamin Young: https://schema.org/CovidTestingFacility - maybe build stuff with this to help folks :)

Benjamin Young: shows how JSON-LD and CSVW can be useful.

Benjamin Young: twitters about it https://twitter.com/westurner/status/1239938629310410752

5. Adjourn

6. Resolutions