Meeting minutes
<LarsG> this is DXWG CNEG Subgroup
<LarsG> rrsagent please draft minutes v2
approve last meeting's minutes
+0
<pchampin> +0
<LarsG> PROPOSED: Approve minutes at https://
<LarsG> +1
<YoucTagh> +1
RESOLUTION: Approve minutes at https://
Updates
roba: talking to OAS (OpenAPI) - will be exploring ConnegP binding to OpenAPI
… OGC will be engagigng OAS and exploring relationship between OAS and ConnecgP
LarsG: update on IETF
… with Reuben Verbough resumed work on IETF draft
… RFC9110 implications being explored
… incorporating editorial comments from @YouicTagh
YoucTagh: explained active vs reactive content negotiation - whether server makes choice or not
LarsG: might replace hints from current draft:
roba: does this apply only to HTTP - does it affect QSA binding?
… how should should we proceed?
LarsG: new information - will need to be decided on - timeframe to be determined
<LarsG> issue is ProfileNegotiation/
LarsG: group to review after initial analysis
roba: Anything from RDF* ?
pchampin: not yet ready - will ping group again to elicit input
… open action to describe "conformance levels" - relationship to profiles
Issues due for closing
<LarsG> https://
roba: LarsG to add links to issue w3c/
roba: to review changes to PR 42
... I'll take 42 offline
<roba> w3c/
roba: I need to add my review
… straightforward
… but it's still blocked
pchampin: maybe nick's review was before some other changes? no
antoine: maybe it's because nick's alias is not in the group
pchampin: nick's review is grey, so not counted
… yes he's not counted as participants of the repo
… I'm assuming he's not listed in the DXWG anymore
… I'm happy to approve the PR
roba: ok, merging
<roba> w3c/
roba: straightforward
pchampin: approving
<roba> w3c/
roba: PR from YoucTagh related to changes in examples from canonical to alternate
YoucTagh: there's an issue on it
LarsG: example 19 shows two canonical links
… I'm not sure about the resolution
roba: yes it could be ok to say that there's one canonical html and one canonical turtle
… maybe we need to change the text. Or we allow canonical to be for a combination of mime-type and profile
LarsG: I think we much change the text, as YoucTagh is right
roba: YoucTagh are you ok with changing the text to say that canonical is for a mime-type?
YoucTagh: it clarifies, yes
<roba> Current text "The default representation – the one that will be returned when no specific representation is requested – will be identified by rel="canonical", other representations by rel="alternate"."
<roba> Proposed: The default representation – the one that will be returned when no specific representation is requested for a specific MIME type – will be identified by rel="canonical", other representations by rel="alternate".
pchampin: is it legal in the first place to have several canonicals?
<pchampin> https://
pchampin: there is a *guideline* advising to have only one, but it has no MUST not SHOULD
<roba> RFC 6596: "To better ensure that applications properly handle the canonical link relation, administrators ought to consider the following guidelines: o Specify only one canonical link relation for a resource. (It would be confusing to consider/label/designate more than one IRI as authoritative.)"
pchampin: we would be stretching the guideline
… or we could remove it (having just one canonical) but then we remove expressivity.
LarsG: the text says, this is what to be returned if there's no indication of media type nor profile
roba: the issue is that if the client says they want html and there's two representations, then there's no way to indicate the prefered one.
LarsG: yes we would lose expressivity
roba: I don't see anything in the IANA register that would provide alternative semantics
antoine: could we keep the current example, but flagging the issue that we've got a specific interpretation of a resource here?
roba: we could
pchampin: RFC doesn't seem to address content negotiation. It addresses "duplicative content"
roba: the profile defines the content
pchampin: any IRI that returns slightly different content. I'm not sure it would fit
… canonical would be appropriate for resources with the same content
… The rel can use several links. Maybe having both 'alternate' and 'canonical' would be ok
… I don't know how clients work with it
<pchampin> https://
roba: I've tried to force bookmarking specific resources using 'canonical' but it failed.
pchampin: the signposting website could be useful for inspiration
LarsG: search engines would use the canonical link to redirect users to
<pchampin> https://
LarsG: this is the only use I've seen so far
… I agree with YoucTagh that we might stretch the RFC too far
roba: we could (1) remove 'canonical' altogether and let implementers do their stuff; (2) add a note about interpretation; (3) mint another link relationship
… we could prepare this and take feedback
… maybe we should take an action to see if having a new link relation type would be realistic
… I have no experience on how registration of new types work
pchampin: there is a template that we can use in the W3C document and the W3C team liaises with IETF
roba: given where we are, we could explore this.
LarsG: RFC8288 has the procedure
<LarsG> https://
roba: we should have a candidate name
ACTION: group to explore options for the link relation type
<YoucTagh> +1
ACTION: roba to cleanup YouTagh 's PR
LarsG: should we close the due for closing?
roba: I'd prefer to clean the PR(s) first
+1
<LarsG> +1
<roba> +1