Meeting minutes
Announcements and Introductions
anatoly-scherbakov: I wanted to summarize a mail thread including Benjamin, David & me
… regarding the PyLD library.
… A belgium organization is working on RDFlib, considering pluging PyLD in RDFlib.
… They are hesitant because PyLD has not been maintained recently.
… If Digital Bazaar is not interested in continuing maintaining it,
… they could host it as an organization (under the umbrella of their RDFlib project)
… or join the project as maintainer.
bigbluehat: that's great news. I'll reach out to Digital Bazaar, and come back to everyone.
… PyLD needs a better home. It is not a priority for DB right now.
… This looks like a good community for that.
frenchrh_: we use Python a lot in our group, by not PyLD.
niklasl: I think this is very promising.
… I tried to plug PyLD into RDFlib a while ago.
<niklasl> niklasl/
niklasl: I don't know the state of PyLD. I became a victim of NIH, so I use my own...
… This was also a good way to understand the algorihms, which I transcribed literally.
… I would not recommend its use for production, but it is licensed as 0BSD so copy/reuse at will.
<niklasl> License 0BSD
frenchrh_: re. RDFlib vs PyLD
… we are a big research group (~60 people)
… depending on the field people are from, they use R or Python
… RDFlib in Python works the same way as in R, which is nice
… we need reliable packages, we build big systems
JSON-LD Recharter Status
bigbluehat: we (Ivan mostly) addressed some small issues
… it has been quiet in the past week
pchampin: the only missing review is from the Privacy group
… Ivan reached out about the status and to get an ETA
… I would be tempted to call for a time-out if we don't get it soon
… we just describe the format, not what data
… and we already have a privacy section
… but there's nothing really critical in it
… I'll check on the status, but I don't think it's blocking
… The larger issue is that we need a redundant Chair
… and it was recommended that we solve that before this goes to the membership
bigbluehat: I have some update about the chair, in the next agenda item.
… +1 to declare a timeout on the privacy review.
… If anything, the new features we are planning to add (caching in document loader) will *enhance* privacy.
New Co-Chair Proposal
bigbluehat: we briefly had Gregg as co-chair, while even then was ill-advice, as Gregg was already chief-editor of most specs
… we don't want to reproduce that
… [introducing Victor Lu]
… Victor is new to the W3C process, but pchampin and ivan confirmed that this is not blocking, they will help him learn the ropes
… we need him mostly as redundancy
… he is +1 on that
pchampin: thanks! that's great news
pchampin: his LinkedIn says he's at Oracle
bigbluehat: but he is not anymore
… he is currently unaffiliated
pchampin: we can have him as an Invited Expert, no problem
… it would be worse if he was affiliated with a non-member organization
Gregg Kellogg Memorial
bigbluehat: we started an email thread a couple weeks ago,
… proposing a eulogy for Gregg to be included in the spec
… TallTed made some improvements
… once I incorporate them, we will make a PR on the json-ld-syntax spec
… we can then decide if we want to include them in the other JSON-LD specs
… we can keep this JSON-LD specific or mention his work more broadly
… we can discuss this in the PR when it is open
pchampin: thanks for initiating that
… we can refine it on the PR, indeed
… I'm in favor of it being on the 3 specs and the 1.1 published specs
<anatoly-scherbakov> +1
pchampin: which I think makes sense as those would not be the same had he not been such an amazing contributor and editor
bigbluehat: the acknowledgement section is "robot-populated"
… it should not be too hard to customize it
<niklasl> +1
bigbluehat: as an aside, there has been some discussion about a kind of larger memorial
… this should be done soon, but unfortunately I don't have the cycles
Open Discussion
<niklasl> Big thanks for moving forward on it.
bigbluehat: any other topic?
… my current position is that we will soon be rechartered
pchampin: I wanted to mention the email that we from Sebastien with Web of Things (WoT)
… we were going to have a joint meeting at TPAC
… but since we are cancelled our meeting due to too few of us attending, we are also cancelling the joint meeting
… interacting with the WoT group has always been interesting
… so, I don't think we'll do a joint meeting at TPAC due to timezones
… but if there is enough interest for interacting with them, we can schedule something
bigbluehat: we left it as you describe: most of us will not be there, and the timeslot in Japan is not easy
… but most people in the WoT CG are in Europe, so we could easily organize a joint meeting at another moment
… they could present what they do, what they need from JSON-LD
… their perspective has always been interesting: they use JSON-LD to describe IoT things
… those things themselves don't do a lot with compaction/extraction, but their description are aggregated in other systems
… they lean heavilty on the graph nature of JSON-LD, as well as inference
… If you are all ok with it, I will arrange the calendaring, probably post-TPAC
anatoly-scherbakov: talking with the WoT group would be interesting.
<anatoly-scherbakov> ack
anatoly-scherbakov: Re. TPAC, is there going to be anything about Gregg? There will be a lot of people from W3C, including from LD.
bigbluehat: I like the idea. We can reach out to the organizers of the major evening.
… not a long thing, but a short memorial in front of everyone.
pchampin: I'm traveling soon, but I will help get this started over email
… the conference I'll be at before TPAC is the Semantic Web Conference
… and we're doing a tutorial there and will be giving a tribute to Gregg before that
… the tutorial is on Sunday, the first day of the conference.
bigbluehat: dlehn, we talked about PyLD at the start of the call
… I'm going to make a pitch internally to suggest we hand it over to the RDFlib group
dlehn: yes, I was the meeting notes
… I got the email about that too
… I don't know what to do with it. I wish I had more time to work on it. It's a change from JS.
… But we have not many use-cases in Python, so this is not a priority
… One of the thing I would like to do if I find time is to ensure all the tests work again.
pchampin: I'll have a look at that Erlang link :)
… I'm also a fan of Rust
… and making a Python binding for a Rust library is relatively easy
… and oxigraph is a Rust library which might be interesting to incorporate
… not sure it does JSON-LD, but there are other Rust libraries which do
… though multiple implementations are good--so keeping PyLD around has it's own value there
dlehn: is that any different than a C++ bound library?
pchampin: the developer experience is much nicer
… I was pleasantly surprised by how easy it is
anatoly-scherbakov: I second the point about Rust bindings, I suggested that in the email thread as an alternative
… I also suggested niklasl's library
… I concur that PyO3 to generate Python bindings for Rust library is really easy
… it has also benefits in term of performance
bigbluehat: if any of you mentioned something that's not on the json-ld website, send me a PR
niklasl: I wanted to ask about some discussions in W3C about how to specify algorithm is a kind of pseudo-language
bigbluehat: yes, this rings a painfull bell
… there was discussion about the relevance of using WebIDL and/or INFRA in our specification, none of which is entirely satisfying
… it is a non-trivial thing for us to update
… I would need a very dedicated editor to refactor this, with a lot of time on their hand
… TRLD is transpiling into Java and JS, I fiddled with the idea of transpiling it to spec text
… transcribing the JSON-LD algorithms to TRLD may be feasible
… On another topic, I recommend Oxigraph, PyOxigraph (a dedicated python binding) and OxRDFlib (RDFlib adapter of the former)
… there are some issues, but Thomas knows about them
… I also second the benefit of pure python implementations
anatoly-scherbakov: did we ever thought of writing specs as linked data instead of free-form?
… i'm not an ontologist, I don't know the answer. I suppose this would be a huge graph.
… Was it ever considered?
niklasl: I have thought of that, it was not my most healthy thought!
… I do have a started for implementing a Turtle parser in Turtle.
… I might put that in a gist.
… The algorithm description is currently very imperative.
anatoly-scherbakov: I agree, and that makes the description verbose.
… Using a functional paradigm would make it more concise.
bigbluehat: anatoly-scherbakov if you have any interest in getting an understanding of these algorithms,
… we would be interested in your lights in moving away from INFRA (which was initially a request from the TAG)
… More of us will have to contribute to whatever is next, it is worth a fresh set of eyes.
niklasl: I will spend from time this week on that idea.
… Generatin INFRA from Python does not seem so hard.
bigbluehat: part of the argument was to be able to reuse the algorithms into other specifications
niklasl: Claude should be able to do that as well
bigbluehat: yes, the challenge would then be to determine whether it is correct
<pchampin> my regrets for the meeting in 2 weeks
bigbluehat: we'll see how the chartering progress and maybe have a meeting