Linked Data Platform (LDP) Working Group Teleconference -- 15 Dec 2014

<trackbot> Date: 15 December 2014

<deiu> Sandro said he's going to be late 20mins and asks if we can postpone LD Patch until he gets there

<scribe> scribe: Alexandre

<scribe> scribenick: betehess

Arnaud: approving the minutes of previous meeting
... no objection: approved
... next meeting could be Jan 6th
... but people could be returning to work that day

<azaroth> +1 to 12th

<azaroth> Otherwise regrets for for the 5th

<SteveS> +1 to 12th

Arnaud: or 12th?

Ashok: 12th!

Arnaud: let's do Jan 12th
... next meeting on Jan 12th
... tracking of actions and issues
... Alex had one action
... we can close it
... we'll talk about the issue later
... let's talk about LDP and Paging
... LDP going to CR, Paging to PR
... there was discussion about the abstract in LDP
... SteveS took an action
... ericP worked on getting the spec out

ericP: choose Jan 16th for the end of PR and CR
... not sure how long the waiting period should be
... for PR, there is a minimum, 3weeks/1month?

Arnaud: end of PR for LDP: minimum is 4 weeks
... on track to be publish on Dec 16th
... then we have the waiting period
... for Paging, as we have no commitment for implementations, so we're planning to wait longer
... so could be 3 months
... Mar 16th sounds more reasonable
... what do others think?

SteveS: day doesn't matter too much to me
... so March sounds reasonable

PROPOSAL: Mar 16th for the end of CR Paging

<deiu> +1

<TallTed> +1

Arnaud: all +1s say that people will be implementing the spec: yay!

APPROVED: Mar 16th for the end of CR Paging

ericP: the LDP ontology uses the LDP paging namespace
... we may need different ns for paging
... eg ldp-paging
... people would have to include both namespaces
... but much more modular
... and no need to modify things after LDP got to Rec

azaroth: the expectation is that LDP Paging wouldn't be used without LDP? And hence modularity would only be one way?

Arnaud: slight overhead with the 2 namespaces
... but kinda makes sense
... and people could come up with a different mechanism for paging, and another ns

SteveS: we talked about that some time, and we decided to keep it in ldp ns, can't remember when or why
... there will always be new ns, how do be layer new terms into LDP then?
... so no -1 ldp-paging
... but like the idea of having one common vocabulary
... and be clear about the status

ericP: in principle, after ldp fgoes to Rec, I weould be permitted to change the terms used in ldp
... but I think I could edit the doc (I guess)
... value is: there is simplicity in having things in one place

Arnaud: look at schema.org
... it's all in there

deiu: there is a performance issue too
... eg tabulator would dereference all vocabularies
... so dereferencing things is costly for us

ericP: it's modularity vs simplicity

<TallTed> what's the URI for the (current) Vocabulary Status Ontology?

<SteveS> http://www.w3.org/2003/06/sw-vocab-status/note

sandro: not strong advocate of 1 namespace
... people use the wrong namespaces all the time

betehess: one namespace means ontology clashing

ericP: what's the title and metadata for the document?
... [enumerating]

<Arnaud> STRAWPOLL: a) keep one namespace, b) move paging into its own namespace

sandro: LDP Paging is part of LDP
... and we'd have to come back to LC if we want to change the NS

ericP: not sure
... if the implementations are not impacted...

sandro: should ask the director
... question is: do you affect somebody

Arnaud: I suggest we keep it as it is: one namespace
... any objection?

<deiu> +0 (mainly because of having to go back to LC)

<betehess> -0.9

Arnaud: I hear no objection
... that settles it

betehess: wait, we should ask the rest of w3c staff

I would prefer having the strawpoll

<deiu> me too

so that we can present it to w3c

Arnaud: don't think ericP convinced anybody (but Alex)

ericP: should I change all the ldp related terms to "stable"
... and leave the paging ones as unstable?

[chorus]: yes

Arnaud: ok, let's move on
... should be published tomorrow
... let's move to LDP Patch then
... betehess had the action item to ask timbl
... hard vs soft delete
... which was having different operations

<deiu> betehess: once thing I didn't make clear in my email, the names for the operations were not discussed, so we can replace them if people come up with better ones

<bblfish> can't hear sandro

sandro: don't think timbl had considered the rollbacxk problem
... b/c he doesn't use it
... he doesn't want to implement a rollback system

<deiu> sandro: Tim didn't really think about the possibility of having rollbacks

<deiu> ... he thinks that you must not send a patch that can result in a rollback

sandro: there should be no rollback

<deiu> ... and he thinks that people should not have to implement systems that do rollbacks

<deiu> ... I too would have a problem with implementing rollbacks

Arnaud: not sure we're why we're discussing about rollbacks

<deiu> Arnaud: I don't really understand how we got to the rollback discussion

<deiu> betehess: if there's a problem with the patch, the resource is left unchanged

<pchampin> Bind and UpdateList can alreday fail

<deiu> Arnaud: maybe the rollback term is too strong

<deiu> ... you cannot have a patch that "kind of" works but fails at the same time

Arnaud: PATCH either succeeds of fails, no in between

sandro: so during the patch, you are not modifying the db
... you are cumulating the add/delete

<deiu> sandro: ok, so as you're applying the patch, you're not modifying the database, and in the processing of the patch you're making the necessary checks, so in that case the "rollback" is trivial

sandro: did I get it right?
... ok, should be fine

<deiu> Arnaud: the preference is therefore to have 4 operations (2 additions and 2 deletes)

Arnaud: so, the preference is to have 4 operations? add/delete can-fail/never-fails?

<deiu> ... are we good with that now?

<deiu> ... is there anything else the editors need at this point so we can close the issue-103?

PROPOSAL: close ISSUE-103 with having 4 operations: add/delete can-fail/never-fails

<pchampin> +1

<deiu> +1

<SteveS> +1

<TallTed> +1

<Ashok> 1

<ericP> sandro, +1

RESOLUTION: close ISSUE-103 with having 4 operations: add/delete can-fail/never-fails

Arnaud: current names are Add AddNew Delete DeleteAny

Arnaud+Sandro: DeleteAny sounds funny to me

scribe: AddNew looks ok

pchampin: share concerns re: DeleteAny
... would prefer Add and Delete would be in same category (can-fail)
... would be my preference

Arnaud: there is an asymmetry

sandro: I'd like that (no assymetry)

<pchampin> pchampin : and the others could be AddNew and DeleteExisting

<TallTed> +1 AddNew/DeleteExisting could fail ; Add/Delete don't fail

<azaroth> +1

PROPOSAL: AddNew/DeleteExisting could fail ; Add/Delete don't fail

<azaroth> +1

<deiu> +1

<SteveS> +1

<Ashok> DeleteIfThere

<pchampin> +1

<deiu> sandro: +1

<TallTed> +1

<deiu> DeleteIfThereAndETagMatches

<TallTed> DeleteIfExists better than DeleteIfThere

<Ashok> One more letter!

Arnaud: no real objections so far?

RESOLUTION: AddNew/DeleteExisting could fail ; Add/Delete don't fail

<pchampin> Sandro, you are right: "ifX" sounds like something that would not fail

Arnaud: any other issue we should be aware of?
... I believe we closed everything
... so when can we published it as CR? (new process track: combined with LC)
... same burden
... need to keep track of public comments
... suggesting we go to CR
... are we ready for such a decision? or do we need more time for people to review the spec?

sandro: and we have to plan the transition meetings
... so the question is only "when"?

Arnaud: believe so

<Arnaud> PROPOSED: Publish LD Patch Format as Candidate Recommendation/Last Call WD

Arnaud: proposing we are doing it now

<pchampin> question re. process: can we make slight editorial changes if we go to CR?

<bblfish> I have not read it

sandro: has anyone other that the editors read the draft?

betehess: timbl read it

Arnaud: what's the exit criteria?

<SteveS> I have not read it since updated but don’t object to move it forward, so many documents I need to patch

sandro: I'd say 2 implementations for the entire test suite

bblfish: can I add metadata to the PATCH?

deiu: it's not using an rdf representation

bblfish: you won't be able to do event-sourcing

betehess: that is orthogonal
... this is just HTTP PATCH

bblfish: want to keep history of changes
... would be cool to have that in the PATCH format
... to know the reason

sandro: "I would like to express my PATCH as a trig document"
... the PATCH could be RDF with a String for the PATCH inside it

<bblfish> ok, I can read it

<SteveS> I can update my review

Arnaud: I am asking for volunteers to read the spec during the holiday

<MiguelAraCo> I can commit to read it

Arnaud: let's make sure that for next call, people have read the spec
... also, would be nice to have a link to a complete test suite

sandro: and the implementation report

<Arnaud> PROPOSED: Publish LD Patch Format as Candidate Recommendation/Last Call WD

<deiu> +1

<pchampin> +1

<Ashok> +1

<SteveS> +1

<TallTed> +1

we can't publish tomorrow (too late), so that'd be Thursday anyway

<Arnaud> RESOLVED: Publish LD Patch Format as Candidate Recommendation/Last Call WD

Arnaud: thank you
... thank you all for joining
... happy holidays!

<bblfish> happy holidays.

adjourned

- DRAFT -

Linked Data Platform (LDP) Working Group Teleconference

15 Dec 2014

Attendees

Contents

Summary of Action Items

Scribe.perl diagnostic output