Dataset Exchange Working Group Teleconference

09 April 2019

Meeting minutes

apologies: makx, LarsG

<roba> +0

<DaveBrowning> +1

proposed: Accept minutes from last meeting

<annette_g> they look empty

<PWinstanley> +1

<AndreaPerego> https://‌www.w3.org/‌2019/‌04/‌02-dxwg-minutes

<AndreaPerego> 0 (was not there)

<riccardoAlbertoni> +1


<annette_g> 0 (was not there)

Resolved: Accept minutes from last meeting

Japan meeting

PWinstanley: we must book by the 12th

PWinstanley: perhaps only 2 people will attend

<AndreaPerego> I am very unlikely to be able to come - but can join remotely.

PWinstanley: are we missing anyone?

<roba> i wont be able to make it.

<AndreaPerego> Do we have a page for the f2f?

I will not be able to attend in person

PWinstanley: My gut feeling is that this is not a "goer" and we should work out what issues should be discussed here and do a virtual meeting

<AndreaPerego> "Virtual" f2f would be good - maybe more than 1

PWinstanley: are there any other views?

PWinstanley: are you happy if I ignore the deadline (not booking)?

<AndreaPerego> Can we try and do a poll?

I am ok with no booking

<AndreaPerego> I mean, before no booking.

proposed: do not book a room

<AndreaPerego> Ok


<riccardoAlbertoni> +1

<DaveBrowning> +1

<roba> +1

AndreaPerego: we should ask the mailing list this

PWinstanley: this meeting seems not to go

Action: PWinstanley to email the mailing list about not having a Fukuoka meetig

<trackbot> Created ACTION-321 - Email the mailing list about not having a fukuoka meetig [on Peter Winstanley - due 2019-04-16].

action items

proposed: close action 319

<roba> yep UTC 2000

<AndreaPerego> +1

<AndreaPerego> close action-319

<trackbot> Closed action-319.

PWinstanley: the spreadsheet is there as a reminder

subgroup progress reports

<roba> * just found my response to conneg meeting time bounced - sent from wrong address

PWinstanley: would like to discuss conneg first and a vote on 2PWD is needed

<PWinstanley> ncar: we had a couple of critical things to address whchi has resulted in 2 small editorial PRs

<PWinstanley> ... and some technical errors have been resolved.

<AndreaPerego> https://‌w3c.github.io/‌dxwg/‌conneg-by-ap/

<PWinstanley> ... There are a large set of small issues to address, as well as some philosophical issues

<PWinstanley> ... we would like to go to 2PWD now

<PWinstanley> the philosophical concern is whether some issues are normative or not

<roba> the spec does not make the QSA implementation normative, either as an option or a specific implementation.

roba: the QSA approach is about human-readablity and is not normative - a recommendation - but it has the conceptual model in common with the HTTP realization

PWinstanley: is all of the contentious material well described? We must put the main points at the front

roba: Yes, but are the issues clarified, is there a mechanism to process the issue, this is the main point

roba: can we improve the clarity of the spec

PWinstanley: we still have this issue of tokens

roba: anyone useing a QSA approach will (likely) use tokens, but must provide URIs

annette_g: questioning the normative nature of the document- one doesn't need to do both realizations, this is clear, but the text about QSA says its normative and I don't think it should be

annette_g: I have written about this in GitHub and the main problem I see is we have a Use Case about human profile selection and I don't think they need to go through a negotiation approach to do this

annette_g: show them what their options are (a list) but I don'd see a Use Case for negotiation as this makes it harder for humens - hides options

<roba> sense is quaified by the following sentence - so can we improve this to avoid this confusion?: "this realization is fully specified here and this document is considered normative for the QSA realization. This realization does not preclude other QSA specifications for profile and content negotiation."

PWinstanley: I can see how this would work for a single server but how about for federated? Would giving options in this scenario were not all servers can do the same thing would work?

annette_g: there would be an application that gives options within a viewport

annette_g: specifying viewport (forms) is beyond the scope of this document

<roba> form based approach is supported by profiles vocabulary - and conneg does not dictate this

<roba> specifically conneg does not go anywhere near telling clients how to handle multiple dimensions - it does not force a cartesian product of choices to be listed

roba: we can't specify how all conneg options are displayed to users. We can't expect implementers to all move to new HTTP methods - too hard. We aren't violating the principles of the web by specifying visible content- use of PROF might guide that. Providing a minimal mechanism for human profile negotiation is the goal - a real set of Use Cases

PWinstanley: is your point that this approach is privileged by its mention above others?

annette_g: yes, there are lots of ways people could implement this and implementer needs may vary

<roba> only two calls...

annette_g: implementers should not be told specifically how to implement

<PWinstanley> ncar: there is a section (key naming) that describes how negotiation has to take place, and it specifically says that there are alternative approaches

<PWinstanley> ... there are only 3 functions involved

<roba> 2

<PWinstanley> ... it is a minimal list

annette_g: I find it odd (confusing) that the document has a normative section then with qualifiers later

<roba> sounds like we need to revisit the communication more than the content...

AndreaPerego: I see a number of things to be fixed - empty boxes to be filled, some with people's names

AndreaPerego: a section (related work) on CSW should be added as it's common, I can draft text for this, similar to OAI-PMH

* we accept your offer!

<PWinstanley> ncar: I can draft the CSW section today, if that was seen as critical, esp if AndreaPerego gave pointers to this and the empty boxes

<roba> i dont see empty boxes...

<AndreaPerego> https://‌w3c.github.io/‌dxwg/‌conneg-by-ap/#related-oai-pmh

<roba> is this a javascript issue at your end?

<roba> QSA discussion is an open issue still - thats the way we deal with these things - lets see if there are more comments on this

<AndreaPerego> s/we'll/I'll/

PWinstanley: does the group feel it's in a position to vote knowing Andrea has raised small editorial issues and annette_g has raised a larger issue?

<roba> issue is already there

AndreaPerego: a suggestion- to include a note in the QSA section about whether this is normative or not is under discussion

annette_g: there are already a lot of those, we may be better off discussing in person before a PWD. I worry about change later

PWinstanley: can others contribute please?

riccardoAlbertoni: my understanding is that we want the 2nd draft out to concerntrate on other deliverables so external comment can be received

<roba> we have one comment that has been identified as being about interpretation of what normative for an optional part means - and one person not seeing a Use Case does no invalidate it when consensus has accepted that Use Case. we need to find a way to move on.

PWinstanley: decision chairs took was that Profiles Guidance doc was not in as mature a shape. Conneg was to be addressed but not if it slowes down DCAT work

<annette_g> there are others. Antoine had reservations as well, for one.

DaveBrowning: I'm not convinced that we can't move this in parallel to DCAT

I agree - different people usually

DaveBrowning: there's a lump that will come through when we come to PR time but seemingly no block now addressing Andrea's issues and perhaps Annette's more complex one. We should not stope trying to fix this

PWinstanley: do you (Dave) have any ideas about addressing these issues

<roba> -1

<annette_g> I would not be screaming

roba: this ability to support human- & machine-readable profile choices for profiles is the my (OGC) motivation here

PWinstanley: perhaps the editors are too close to a specific set of mechanisms?

roba: actually this is a process issue about which Use Cases apply and we shouldn't block on not everyone buying in to all use cases

roba: we may also have a communication issue which can be improved

PWinstanley: I understand this but I am exporing other options

DaveBrowning: point on dropping QSA section- our charter says we must provide fallback mechanisms to HTTP

annette_g: I am not the only one with reservations about QSA, Antoine & Alejandra too

annette_g: for consensus, we all need to listen to one another

<roba> alejandra has participated so i dont think there is an issue here - we dont have other active discussions

PWinstanley: what would the document look like if you (Annette) were creating it? How would you address QSA?

<roba> thtas what it is - so please feel free to suggest better introductory wording

annette_g: if the charter really requires this (fallback approaches) then I would consider a non-normative section with an example of dataset selections via profile

annette_g: issued as you would an Open Source software package - shared code. Don't understand the need to make it a standard

PWinstanley: we are at the end of today with discussion still in place.

<DaveBrowning> DaveBrowning: Charter text at https://‌www.w3.org/‌2017/‌dxwg/‌charter#deliverables. "An explanation of how to implement the expected RFC and suitable fallback mechanisms as discussed at the SDSVoc workshop."

PWinstanley: do you (annette_g) understand Antoine & Alejandra's view?

annette_g: Antoine's yes

<DaveBrowning> DaveBrowning: Though doesn't mean QSA specifically...

PWinstanley: I would like to put this out to wider discussion with points of contention known. This may allow others (feedback) to assist us with these issues

PWinstanley: we have another meeting before Easter break, do you (annette_g) think we can push a vote to next week - there are things we can do in the interveaning week for this?

PWinstanley: can I ask annette_g to Articulate what would bring about a more balanced view on QSA

annette_g: I have made my points in issues - what more can I do?

PWinstanley: we need to have issues in writing (in the document?), not just what we agree in discussions (meetings)

annette_g: if you want something quick, pull that section out, no time to re-write in a week

PWinstanley: we will have to deal with this in correspondence

PWinstanley: we are at an impass but need text that suits everybody, can you (annette_g) provide a little text, rather than pulling a section out?

PWinstanley: if we can work on the text and show this (QSA) is one of a range of potential approaches, this might be more suitable to Annette, Antoine & Alejandra's view

annette_g: can I work on the text of a subset?

PWinstanley: yes

Action: annette_g to Work on the text of a subset of conneg to better convey her (and others) view

<trackbot> Created ACTION-322 - Work on the text of a subset of conneg to better convey her (and others) view [on Annette Greiner - due 2019-04-16].

<riccardoAlbertoni> thanks good night

<AndreaPerego> Thanks, bye!

Summary of action items

  1. PWinstanley to email the mailing list about not having a Fukuoka meetig
  2. annette_g to Work on the text of a subset of conneg to better convey her (and others) view

Summary of resolutions

  1. Accept minutes from last meeting
Minutes manually created (not a transcript), formatted by Bert Bos's scribe.perl version 2.49 (2018/09/19 15:29:32), a reimplementation of David Booth's scribe.perl. See CVS log.


Succeeded: s/regrets: makx, LarsG//

Succeeded: s/regrets+: Antoine, Alejandra/regrets+ Antoine, Alejandra/

Succeeded: s/ack: DaveBrowning//

Succeeded: s/topic: outstanding actions//

Succeeded: s/+/+1/

Succeeded: s/privilidged/privileged/

Succeeded: s/minumal/minimal/

Succeeded: s/woould/would/

Succeeded: s/specificlally/specifically

Failed: s/we'll/I'll/

Succeeded: s/notmative/normative/