W3C

- DRAFT -

Linked Data Platform (LDP) Working Group Teleconference

17 Dec 2012

See also: IRC log

Attendees

Present
+329331aaaa, ericP, Ruben, antonis, deiu, Arnaud, SteveBattle, MacTed, bblfish, Ashok_Malhotra, SteveS, Yves, AndyS, raul, krp, rogerm, nmihindu, +1.631.444.aabb, Kalpa
Regrets
Chair
SV_MEETING_CHAIR
Scribe
MacTed

Contents


<trackbot> Date: 17 December 2012

<Ruben> who's talking now?

ericP - is totally broken up

<bblfish> hi

<Ashok> Hi

<scribe> scribenick: MacTed

<stevebattle2> +1

<SteveS> +1

<antonis> +1

<bblfish> +1

PROPOSED: approve minutes of 2012-12-10

<Arnaud> +1

<deiu> +1

RESOLUTION: approve minutes of 2012-12-10

Open Actions

<SteveS> action-17?

<trackbot> ACTION-17 -- Steve Battle to prepare first draft of the Use Cases and Requirements Document -- due 2012-10-29 -- OPEN

<trackbot> http://www.w3.org/2012/ldp/track/actions/17

action-17?

<trackbot> ACTION-17 -- Steve Battle to prepare first draft of the Use Cases and Requirements Document -- due 2012-10-29 -- OPEN

<trackbot> http://www.w3.org/2012/ldp/track/actions/17

<bblfish> we are looking at open actions http://www.w3.org/2012/ldp/track/issues/open

close action-17

<trackbot> ACTION-17 Prepare first draft of the Use Cases and Requirements Document closed

Issues Pending Review

<bblfish> looking at issues: http://www.w3.org/2012/ldp/track/issues/

<bblfish> especially http://www.w3.org/2012/ldp/track/issues/pendingreview

issue-41?

<trackbot> ISSUE-41 -- Standard way to manage members with attachments -- pending review

<trackbot> http://www.w3.org/2012/ldp/track/issues/41

FPWD of UCR - http://www.w3.org/2012/ldp/wiki/Use_Cases_And_Requirements

<SteveS> Updated at http://www.w3.org/2012/ldp/hg/ldp-ucr.html

s/http:\/\/www.w3.org\/2012\/ldp\/wiki\/Use_Cases_And_Requirements/http:\/\/www.w3.org\/2012\/ldp\/hg\/ldp-ucr.html/

( discussion about publication timing, moratoriums, etc. )

<Arnaud> PROPOSED: publish current UCR draft as First Public Working Draft

<SteveS> +1

<stevebattle2> +1

<Yves> +1

<nmihindu> +1

<deiu> +1

<Arnaud> +1

<roger> +1

PROPOSED: publish current UCR draft as First Public Working Draft, http://www.w3.org/2012/ldp/hg/ldp-ucr.html

+1

<bblfish> +1

<Ruben> +1

<AndyS> +1

<Yves> I got confirmation that next round of publication will be jan 3rd

<raul> +1

<ericP> +1

<ericP> note that the SOTD proposed last week indicated that it's not in a final state

RESOLUTION: publish current UCR draft as First Public Working Draft, http://www.w3.org/2012/ldp/hg/ldp-ucr.html

<ericP> last week's proposed SOTD

<ericP> +1 to giving SteveS jurisdiction over the wiki page

SteveS: suggests we remove the deprecated content from wiki

<ericP> ... to corrupt as he sees fit

<bblfish> I think it make sense to remove the content and link to the spec

<ericP> the OWL WG used a schema that Sandro concocted to publish directly from wiki.

<stevebattle2> We don't want the wiki content to drift away from the published content

+1

<ericP> if we're not going to use such a system, and the authors want to work from resepc, yes, avoid confusion by nulling the page

<bblfish> should one have a Proposal?

<Yves> +1 to ericP

PROPOSED: wipe out wiki content of this page, leaving only in hg/respec

<Arnaud> +1

<antonis> +1

<ericP> +1 to Yves's +1 of ericP

<stevebattle2> +1

<deiu> +1

<bblfish> +1

<Yves> +1

<Ruben> +1 ok

<krp> +1

<nmihindu> +1

<ericP> +1

<SteveS> +1 will put "previous version" link to be something like http://www.w3.org/2012/ldp/wiki/index.php?title=Use_Cases_And_Requirements&diff=1678&oldid=1676

RESOLUTION: wipe out wiki content of http://www.w3.org/2012/ldp/wiki/Use_Cases_And_Requirements , leaving only in hg/respec

<stevebattle2> woohoo

editor's draft of LDP spec, http://www.w3.org/2012/ldp/hg/ldp.html

SteveS: LDP spec remains as was...

Discussion of Open Issues

Arnaud: mailing list discussion suggested we clear up some pieces of LDP model before moving to other specific issues
... particular concerns over container modeling
... Lyon F2F concluded there were 2 possible models -- aggregation (weak) and composition (strong)
... composition requires something special from server -- one HTTP DELETE on container means that server must also delete all members
... aggregation requires nothing special from server
... Lyon F2F concluded that spec should only discuss composition model
... if WG members want aggregation model as well, need proposal of such

<bblfish> +1 Arnaud makes a good summary of the composition problem

<stevebattle2> Nice concise summary of the issues, Arnaud

Ashok: has figured out AtomPub model, and believes this is what Erik wants adopted
... for aggregation, you put a link into the container. when the container goes away, the links go away, but the linked resource may remain or go away

Arnaud: doesn't think Erik is saying, do like AtomPub, but -- it would be useful to document LDP Model, and AtomPub documentation may be useful frame to start from
... we don't have to do as AtomPub, but we have to address same questions/concerns

stevebattle2: concurs with Arnaud's description of Erik's position

<ericP> i think a more important discriminator is how much trouble will be caused by deleting resources 'cause they happen to be in containers

bblfish: it's better to start simple if we can. would be useful to know if anyone has implemented LDP as it stands.

<ericP> i think it's the use more than the implementation which will demonstrate the cost vs. benefits of deleting contained resources

Arnaud: at F2F, seemed like people wanted containers so they could do paging, but there are other ways to do paging...

<krp> It was issue-33 http://www.w3.org/2012/ldp/track/issues/33

Ashok: agree that we need to spell it out, whether like AtomPub or otherwise

<roger> +q

stevebattle2: LDP will definitely allow aggregation via RDF. maybe we just need such a statement...

<bblfish> +1 for the idea of having a section how to do aggregation using RDF

MacTed: agrees that statement would be VERY important, as lacking it, it looks like composition is *all* that LDP permits...

<stevebattle2> But I wouldn't want to restrict people's creativity in using RDF in novel ways.

<AndyS> Has anyone tried inserting in to an RDf collection/container over HTTP? !

<ericP> Arnaud, has IBM seen need for deleting resources when deleting containers?

roger: aggregation is important. experience is that it's necessary, and it should be defined in LDP 1.0

<SteveS> ericP from what we do for tool integration, doesn't occur too often. Typically the client deletes each resource on its own, then removes the container.

bblfish: agrees that saying how aggregation is done with pure RDF is important, and then move on to what LDP's composition gives that simple RDF aggregation doesn't

stevebattle2: responding to Ashok's mailing list proposal...
... primary objection is that composition and aggregation are being confused by some people

<bblfish> I also mentioned that the arguments should come from use cases from which we can then argue

stevebattle2: wants a simple composition model, not to undermine that with lots of metadata

<ericP> SteveS, i take that as some evidence that the initial design should be for aggregation

SteveS: would like issue proposals to be separated from agenda threads on mailing list...

<bblfish> I mean arguments for composition of containers should start from a use case and argue why that cannot be implemented without strong containers.

Arnaud: maybe we should have a vote on Ashok's proposal? though I'd hope it would be a more complete formulation, not just a single property

Ashok: if we implement aggregation with members within a container, then we can do as I proposed
... if we implement as AtomPub, then different mechanisms are required
... must decide between members-within-container or pointers-within-container

<ericP> it's always references, like java

<ericP> isn't aggregation what's already in the spec?

<ericP> what's missing?

<stevebattle2> compregation is in the spec

<roger> +q

<ericP> ahh, roger that

<ericP> so an aggregation proposal would be to remove the text that talks about DELETE?

Arnaud: seems to me we always have links within containers... starting from RDF as we are

bblfish: thinks we should have a use case presenting why we need more than RDF-based aggregation, justifying composition model

Arnaud: current spec has mixed handling

SteveS: spec is primarily based on aggregation. implementation may decide what happens with container-members it manages, and how to handle members with remote management...

<AndyS> Precludes linking?

<ericP> and an composition proposal would be to strengthen the DELETE text

<ericP> and a hybrid would include a mechanism to discriminate

<bblfish> bblfish: my last point was that we should write up how to do aggregation by writing a document which is just a collection

ericP: an aggregation proposal would be to remove the text that talks about DELETE; composition proposal would be to strengthen the DELETE text; hybrid would include a mechanism to discriminate

<stevebattle2> It's not a binary choice between the two though

Arnaud: perhaps we need SteveS to revise spec based on discussion to date, before we focus too strongly on it

<bblfish> ok, looks like I should write it up

<stevebattle2> It's business as usual, Roger, using rdfs:member, rdf:List, ...

<SteveS> I plan to write/propose something as well but want to base it off what lands as composition

<SteveS> stevebattle2 but what if you need to create a resource, where do you post it?

<bblfish> roger, should we hook up to see if we agree?

<stevebattle2> "where do you post it?" is a question about composition, not aggregation.

( proposal to come... )

<AndyS> stevebattle -disagree - aggregation needs "add to list"

issue-5?

<trackbot> ISSUE-5 -- Add a section explaining how LDBP is related to Graph Store Protocol -- pending review

<trackbot> http://www.w3.org/2012/ldp/track/issues/5

<stevebattle2> "add to list" can be defined as an RDF patch.

<AndyS> ... how to do that? What is the PATCH format?

<SteveS> stevebattle2 seems odd to have two ways to create

<stevebattle2> Well, changesets, if I ruled the world :)

<AndyS> (both list and seq are hard to do via serialized formats because of bNodes and numbering resp)

<AndyS> changesets can't do it! Do not work on rdf:Lists!

<stevebattle2> sigh...

<AndyS> seq similarly - hard to guess the number of the rdf:_N

Issue-5 section explaining how LDBP is related to Graph Store Protocol, http://www.w3.org/2012/ldp/track/issues/5

Arnaud: thinks we don't need anything in spec about this, given we have note about it on LDP homepage

MacTed: I think the paragraph relating LDP to GSP is necessary within spec

SteveS: wonders whether there are any other specs we need to address in similar fashion?

bblfish: maybe we just set this as postponed, and see whether explanation/connection is easy later?

Arnaud: GSP doesn't require SPARQL, though it's written with it...

<stevebattle2> The operational semantics of GSP are defined in SPARQL.

<ericP> re implementing both, i've implemented all of GSP and a bit of LDP.

Arnaud: interested whether GSP server can also be LDP server, or will there be difficulties in being both -- how difficult is it to support both?

<ericP> the main requirement was that the server remembered the differences between LDP containers and GSP endpoints

<bblfish> you could POSTPONE

Arnaud: issue-5 resolution for now is ... no resolution.

<ericP> other than that, they could peacibly coexist

issue-12?

<trackbot> ISSUE-12 -- Can HTTP PATCH be used for resource creation? -- open

<trackbot> http://www.w3.org/2012/ldp/track/issues/12

<bblfish> I think that is easy

<bblfish> close it

Issue-12 -- Can HTTP PATCH be used for resource creation? -- http://www.w3.org/2012/ldp/track/issues/12

Arnaud: this seems clear --no, you cannot create a resource using PATCH

<stevebattle2> I'd support not using patch for creation - POST is causing enough trouble by itself.

<ericP> pushes the expressivity into the stratosphere

<ericP> how about "PATCH on a container is undefined behavior"?

<roger> ... did we fix the dates for the F2F ?

<ericP> my proposal is the "undefined behavior" text above

<bblfish> mhh, interesting arguments

SteveS: wouldn't prohibit from using PATCH... would recommend using POST and strongly discourage PATCH, but wonders value of prohibition

stevebattle2: first arose because HTTP allows resources to be created with PATCH

<stevebattle2> +1

<Yves> +1

<rgarcia> +1

<SteveS> +1

<Ruben> +1

<antonis> +1

<bblfish> +1

<deiu> +1

<ericP> i prefer to speak in terms of defined behavior vs. discouragement

<Arnaud> PROPOSED: don't prohibit using PATCH... but recommend using POST and strongly discourage PATCH

<Arnaud> +1

<rgarcia> +1

<stevebattle2> +1

<roger> +1

+1

<Ruben> +1

RESOLUTION: don't prohibit using PATCH... but recommend using POST and strongly discourage PATCH

<bblfish> thanks all

<Kalpa> happy holidays and merry christmas all

<stevebattle2> bye

next meeting in January

<deiu> thank you and merry christmas!

<Arnaud> meeting adjourned

<ericP> !

<ericP> tx

<antonis> happy holidays everyone!

<ericP> lol

<ericP> i was asking for a soundcheck

<ericP> she says "happy holidays"

trackbot, end conference

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.137 (CVS log)
$Date: 2012-12-17 16:04:02 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.137  of Date: 2012/09/20 20:19:01  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

WARNING: Bad s/// command: s/http:\/\/www.w3.org\/2012\/ldp\/wiki\/Use_Cases_And_Requirements/http:\/\/www.w3.org\/2012\/ldp\/hg\/ldp-ucr.html/
Succeeded: s/stevebattle2/SteveS/
Succeeded: s/nto/not/
Succeeded: s/bost/both/
Found ScribeNick: MacTed
Inferring Scribes: MacTed
Default Present: +329331aaaa, ericP, Ruben, antonis, deiu, Arnaud, SteveBattle, MacTed, bblfish, Ashok_Malhotra, SteveS, Yves, AndyS, raul, krp, rogerm, nmihindu, +1.631.444.aabb, Kalpa
Present: +329331aaaa ericP Ruben antonis deiu Arnaud SteveBattle MacTed bblfish Ashok_Malhotra SteveS Yves AndyS raul krp rogerm nmihindu +1.631.444.aabb Kalpa

WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth

Found Date: 17 Dec 2012
Guessing minutes URL: http://www.w3.org/2012/12/17-ldp-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]