– DRAFT –
RCH weekly

Scribe (most recent first) markus_sabadello, PhilA, seabass, DLongley, Manu, Gregg, pchampin, Ahmad, AndyS

phila: Given the results from the meeting time poll, I would suggest making it 3PM UK time alternating between Wednesdays and Tuesdays each week

gkellogg: RDF* is doing first public working draft tomorrow

gkellogg: the timing is still off, so we should still target 2.1 for now. Quoted triples we don't have a solution for; we could use what the RDF-star WG produce, but they aren't close to that yet.

seabass: have to apologise for the delay in the PR for issue #4; still intend to do it this week

<gkellogg> w3c/rdf-canon#97

gkellogg: te PR has fairly small changes. It changes the definition of normalized dataset.

markus_sabadello: does this close issue 4?

gkellogg: I think there was some confusion about the ordering of N-Quads; this is currently required in the specification

<Zakim> seabass, you wanted to mention PR for issue 4

<markus_sabadello> seabass: I thought it looks like an improvement, but doesn't fully close issue 4

<phila> seabass: That looks like an improvement, but it doesn't close Issue 4 IMO

<markus_sabadello> seabass: IIRC, ivan didn't think it was appropriate for the sorting to be abstract, but rather a concrete step in serialization

<phila> ... Ivan didn't seem to think it was appropriate to order the abstract dataset

<markus_sabadello> gkellogg: By the time you get to step 7, all the blank nodes in the data set will have been assigned a canonical identifier, this assignment is recorded in the ordered map

<markus_sabadello> gkellogg: If you just look at serialization, it says that when creating the canonical form of a quad, you must use the canonical label from the map

<markus_sabadello> gkellogg: The assignment of the canonical labels is done prior to this step

<markus_sabadello> gkellogg: This provides the opportunity to provide alternative labeling

dlongley: The ordering might matter in the context of issue #89, which is my only apprehension about merging this now

dlongley: What you need to know at that stage is how to map the labels to canonical labels

gkellogg: Are we satisfied that the PR satisfies #4?

<markus_sabadello> seabass: It looks like this satisfy issue 4, but adds extra layers of non normative abstraction, namely moving/sorting to an earlier stage

<Zakim> gkellogg, you wanted to say that the fact the map is ordered may not be important

gkellogg: I don't think it's necessary to have ordering this early, but once you get to step 7, other use-cases might benefit from having an already-normalised, ordered dataset.

markus_sabadello: the primary reason for blank node label normalisation is to support selective disclosure.

<markus_sabadello> gkellogg: Selective disclosure is not something we define, but something we should enable

<Zakim> dlongley, you wanted to propose we merge PR 97 and move onto 89

<gkellogg> +1 to merging #97 and close #4

dlongley: What we're doing in this WG is defining the primitives. I would propose merging #97 and closing #4.

phila: RCH is broader than the VC use cases which is where selective disclosure issue comes from

<dlongley> is fine with leaving it open for now

<dlongley> +1 to leave open and merge soonish -- and move onto the other issues

<dlongley> +1 to trying to keep issues simpler and closed sooner

<markus_sabadello> seabass: Also prefer smaller issues, except maybe in cases like this where the purpose of the specification is being discussed

<markus_sabadello> w3c/rdf-canon#92

markus_sabadello: Let's leave #4 open for a few days; and consider closing #92 at the same time as #4

<dlongley> +1 to markus

+1 to phila's resolution

<markus_sabadello> seabass: Whose commit comes first doesn't seem important. The process shouldn't bog people down.

<markus_sabadello> seabass: I was suggesting the PR could be merged in 2-3 days, because it doesn't mean anything for the group's consensus.

<markus_sabadello> phila: Does it make a difference for gkellogg if the PR is merged at the end of the week, or by next meeting?

gkellogg: there is always the risk of rebases, which can be lossy if one isn't careful

gkellogg: RDF-star WG has PRs open for 1 week or 2 weeks for substantive changes. We can operate more quickly, because we don't have as many responsibilities here

<markus_sabadello> seabass: The algorithm seems to add the concept of an ordered map

<markus_sabadello> seabass: If you solved issue 4, but also added something else that someone may not agree with, would you be upset if the PR gets reverted

markus_sabadello: I think we can always change text more than once

<markus_sabadello> PROPOSAL: Today, mark issues 4 and 92 as pending-close. Merge PR 97 on friday, and close issues 4 and 92. Leave issue 89 open. Create new issue about adding a diagram about the algorithm.

+1

<dlongley> +1

<markus_sabadello> +1

<phila> +1

<gkellogg> +1

<dlehn> +1

RESOLUTION: Today, mark issues 4 and 92 as pending-close. Merge PR 97 on friday, and close issues 4 and 92. Leave issue 89 open. Create new issue about adding a diagram about the algorithm.

gkellogg is a much more magnanimous author than those I have had to deal with in other WGs :)

<markus_sabadello> w3c/rdf-canon#92

<Zakim> dlongley, you wanted to say 89 needs some kind of resolution and may soon cause trouble for VCWG if we don't address it

<markus_sabadello> phila: Meetings will now always be at 10am ET, alternating between Wednesdays and Tuesdays