W3C

RDF-star

12 March 2021

Attendees

Present
AndyS, gatemezing, gkellogg, james, olaf, ora, pchampin, rivettp, TallTed, thomas
Regrets
-
Chair
pchampin
Scribe
ora, pchampin

Meeting minutes

Announcements and newcomers

Daylight Saving Time (US) and Summertime (EU)

pchampin: switch happens at different times in the US and in Europe

pchampin: what would be the reference time zone?

<TallTed> typical practice for w3 work is Boston/New York, but it won't make a lot of difference for me

<TallTed> US would be an hour later (as it would be noon where it is now 11am)

<pchampin> if we stick to UTC: https://www.timeanddate.com/worldclock/fixedtime.html?msg=RDF%2A+call&iso=20210319T16&p1=1440&ah=1

gkellogg: W3C uses Boston time

pchampin: using European time, would be noon for Boston, 9 am for SF

pchampin: gkellogg's argument is valid, though, not to interfere with other W3C meetings

<TallTed> this just impacts the next two weeks, correct?

pchampin: yes

<thomas> gregg has the only argument, so why not take taht?

pchampin: propose to stick with US time

<pchampin> PROPOSED: stick to US time to avoid conflict with other W3C meetings

<pchampin> pchampin: +1

<thomas> +1

<gkellogg> +1

+1

<TallTed> +1

<william> +1

<olaf> +1

<rivettp> +1

<AndyS> +1

<james> +0

gkellogg: strong push in the US to eliminate DST

pchampin: we will stick with the US time

Resolution: stick to US time to avoid conflict with other W3C meetings

pchampin: next week 1 hr earlier for Europeans

SPARQL-eval test suite

<pchampin> PR: https://github.com/w3c/rdf-star/pull/114

andys: no such things as finished test suite

andys: generally good coverage now

andys: syntax tests updated

pchampin: we must reach out to implementors

<gkellogg> My implementations pass all the tests.

andys: I would like to merge now

<pchampin> PROPOSED: merge PR 114 befor advertising the test suite to implementers

<gkellogg> +1

<pchampin> +1

<TallTed> +1

+1

<AndyS> +1

<thomas> +1

<james> +1

<olaf> +1

<rivettp> +1

<william> +

<william> (sorry)

<william> +

<william> +1

Resolution: merge PR 114 befor advertising the test suite to implementers

<william> (numlock was off :-)

gkellogg: we might consider publishing an implementation report early

gkellogg: I have tooling to check

<pchampin> example: https://w3c.github.io/json-ld-api/reports/

andys: if we get good coverage we should publish

pchampin: sounds like a good idea

pchampin: will add a link to the spec

<TallTed> EARL is common (probably best, probably not only) practice for implementation reports these days.

<TallTed> per-test results are optimal for CR->PR->TR, and probably also for CG->Report->WG

Action: gkellogg to make a PR for the implementation report

pchampin: we should make an announcement

andys: could someone check the HTML manifest?

gkellogg: to check that the HTML manifest covers all?

andys: yes, the HTML manifest is generated from Turtle

andys: set of tests does not prove your implementation

pchampin: we should be careful to avoid any statements about completeness

pchampin: this is work in progress

pchampin: raise awareness on the mailing list

pchampin: any other channels?

andys: give people the chance to look before we go to a wider audience

<james> i do not necessarily intend to exercise the test suite before the report is complete

gkellogg: other groups' impl. reports are not necessarily frozen, late entries are accepted

pchampin: nothing is frozen

pchampin: why not use it early?

pchampin: I will email the list

pchampin: there are discussions on the list about function descriptions, should we discuss?

james: we can use the list

Define a URI for the class of embedded triples

pchampin: what namespace?

not ideal to proliferate namespaces

changing an IRI later does not work well

we do not, however, have the authority to change RDF

ora: why is proliferation of namespaces bad?
… easy to create an RDF-star namespace

andys: how much would go to a new namespace?

andys: if it is very few things, might not be worth it

gkellogg: comes down to the question about the purpose of this CG

gkellogg: if we are creating a new spec, makes sense to have a new namespace

gkellogg: if it is to advise other groups, then maybe not

gkellogg: if there is going to be a new RDF WG, then our work is considered an experiment

gkellogg: we can invent new IRIs, we just don't have the authority change RDF, etc.

thomas: semantics not stable

thomas: so perhaps not a good idea to use the RDF namespace

thomas: more prudent to use an RDF* namespace

pchampin: the whole spec is unstable

pchampin: nothing will be fixed until we reach a stable state

thomas: semantics unproven for now

ora: the rdf vs.rdfs issue is historical
… there was 2 different groups
… retrospectiveley, it was unfortunate

andys: missed thomas' point: what is tied to semantics

andys: i am looking at the effect on users

andys: I suggest we say "we propose that..."

gkellogg: RDFa added stuff

ora: RDF namespaces are fixed

gkellogg: any WG is authorized to update namespaces

gkellogg: CG is input WGs

ora: I doubt that it's true that any WG is authorized to change the RDF namespace

gkellogg: we did in the JSON-LD WG, figured this out with Ivan Herman

<TallTed> WGs can change W3 namespaces, *upon approval by W3 Management*, which is not automatic but is typically granted with suitable justification from the WG

gkellogg: we did discuss whether the JSON datatype should be in the RDF NS

ora: feels strange; changing the NS is invalidating the spec

AndyS: this is an addition, not a change

ora: but implementation may enforce the notion that RDF and RDFS namespaces are fixed

TallTed: enforcing something that is not fixed is an implementation error

gkellogg: JSON-LD process was to ask for comments from the community first

gkellogg: very little feedback

gkellogg: no evidence of implementations breaking

thomas: what if semantics change? We define a new term/

gkellogg: RDF 1.1 does not have plain literals

william: Notation3 CG made the decision to reuse old namespaces

william: we did not want to add namespaces

william: TimBL said to go ahead

<pchampin> STRAWPOLL: put our IRI(s) in the RDF namespace

<gkellogg> +1

<pchampin> +1

-1

<rivettp> +1

<james> +1

<thomas> -1

<AndyS> +1

<william> +0

<olaf> +0

<TallTed> +0

no consensus

pchampin: polite thing to do is to seek broader feedback

andys: if something is controversial it is a good idea to make sure people pay attention

andys: (anecdote about John McCarthy)

Summary of action items

  1. gkellogg to make a PR for the implementation report

Summary of resolutions

  1. stick to US time to avoid conflict with other W3C meetings
  2. merge PR 114 befor advertising the test suite to implementers
Minutes manually created (not a transcript), formatted by scribe.perl version 127 (Wed Dec 30 17:39:58 2020 UTC).