IRC log of rdf-wg on 2012-10-29

Timestamps are in UTC.

07:55:20 [RRSAgent]
RRSAgent has joined #rdf-wg
07:55:20 [RRSAgent]
logging to
07:55:32 [sandro]
zakim, dial Rhone_4
07:55:32 [Zakim]
sorry, sandro, I don't know what conference this is
07:55:40 [sandro]
zakim, this will be rdf
07:55:40 [Zakim]
ok, sandro; I see SW_RDFWG(TPACF2F)2:00AM scheduled to start 115 minutes ago
07:56:22 [sandro]
zakim, dial Rhone_4
07:56:22 [Zakim]
ok, sandro; the call is being made
07:56:23 [Zakim]
SW_RDFWG(TPACF2F)2:00AM has now started
07:56:25 [Zakim]
08:31:58 [sandro]
gkellogg, manu1 you're asleep, I imagine.
08:31:59 [sandro]
but we're on the phone.
08:32:28 [ivan]
ivan has joined #rdf-wg
08:34:51 [Arnaud]
Arnaud has joined #rdf-wg
08:40:23 [cygri]
cygri has joined #rdf-wg
08:41:19 [sandro]
RRSAgent, pointer?
08:41:19 [RRSAgent]
08:41:22 [AZ]
AZ has joined #rdf-wg
08:41:25 [sandro]
scribe: eric
08:41:38 [Guus]
Guus has joined #rdf-wg
08:41:47 [davidwood]
davidwood has joined #rdf-wg
08:42:04 [pchampin]
pchampin has joined #rdf-wg
08:42:05 [FabGandon]
FabGandon has joined #rdf-wg
08:42:23 [sandro]
meeting: RDF WG F2F3
08:42:31 [sandro]
08:42:35 [sandro]
chair: David, Guus
08:42:46 [SteveS]
SteveS has joined #rdf-wg
08:42:57 [sandro]
RRSAgent, make record public
08:47:34 [ericP]
topic: intros:
08:47:50 [ericP]
ericP: W3C, HCLS RDF geekery
08:48:05 [ericP]
sandro: RDF geekery, eGov
08:48:22 [ericP]
FabGandon: expertise in KR
08:48:32 [ericP]
cygri: DERI gallway ireland
08:48:42 [ericP]
... databases, RDF sparql
08:49:00 [ericP]
Arnaud: standards guy at IBM, ex W3C staff
08:49:00 [cygri]
08:49:18 [ericP]
pchampin: univ lyon (500m away)
08:49:25 [ericP]
ivan: w3c
08:50:00 [ericP]
FabGandon: ac rep of INRIA, was in this group until falling into an administrative black hole
08:50:35 [ericP]
SteveS: IBM rational, member of LDP, using Rdf for tool integration
08:51:09 [ericP]
eve raymond: BBC, linked data for media industry
08:51:39 [yvesr]
s/eve raymond/yves raimond
08:51:45 [sandro]
Guest: Steve Speicher
08:51:53 [ericP]
@@1: yarcdata, eureka (rdf triple store), observer
08:52:23 [ericP]
davidwood: selling RDF for food
08:52:48 [ericP]
Guus: prof of computer science in web and media at univ amsterdam
08:52:56 [ericP]
... working in cultural heritage
08:53:24 [ericP]
s/@@1/Shoaib Mufti/
08:53:50 [sandro]
Guest: Shaoib Mufti
08:55:34 [ericP]
topic: objectives
08:56:17 [ericP]
davidwood: agenda organized around deliverable documents
08:56:25 [ericP]
-> deliverable documents
08:56:59 [ericP]
davidwood: Concepts, Turtle, JSON-LD are close
08:57:35 [ericP]
Guus, JSON-LD not strictly in the critical path (for charter extension)
08:57:39 [ericP]
Guus: JSON-LD not strictly in the critical path (for charter extension)
08:58:07 [ericP]
sandro: we need to be done with anything controversial
08:59:26 [ericP]
ivan: good if we have core docs at LC (Concepts) or CR (Turtle)
09:00:04 [ericP]
davidwood: some of these docs don't require much update: XML and schema
09:00:09 [ericP]
... get to primer later
09:00:38 [ericP]
ivan: the primer is a note in many WGs
09:01:00 [ericP]
sandro: though it's already a REC in RDF.
09:06:49 [ivan]
s/@@1/Shoaib Mufti/
09:07:21 [pchampin]
pchampin has joined #rdf-wg
09:07:27 [ericP]
cygri: we should discuss a document or sections in other docs about upgrading from RDF 1.0
09:08:13 [ericP]
... should address the fear that the change that 1.0->1.1 breaks stuff
09:10:00 [ericP]
sandro: do we have any reason to believe that danbri will edit RDF Schema? (given that he's changed jobs since volunteering)
09:10:36 [ericP]
davidwood: no, so we need a second
09:13:20 [ericP]
topic: Primer
09:14:13 [ericP]
cygri: re: evolution text, we can point out what's new and talk about what you have to do 'cause simple literals don't exist any more
09:14:36 [AndyS]
AndyS has joined #rdf-wg
09:15:07 [ericP]
ivan: only non-additive change is the simple literal
09:15:35 [ericP]
... until now, if i asked for the datatype for a simple literal, i got back nothing
09:15:41 [sandro]
09:15:52 [ericP]
cygri: impact is low, but you might have to adapt some application code
09:16:07 [ericP]
... spelling that out is worth doing
09:16:22 [ericP]
... e.g. if you use an RDF lib or store, here's what changes for you
09:16:36 [ericP]
... .. if you edit another spec, here's what changes for you
09:16:57 [ericP]
... only a paragraph or a few bullet points, but needs to be included
09:17:18 [ericP]
sandro: OWL created a "New Features and Rationale"
09:17:28 [ericP]
... would be smaller for us.
09:17:56 [ericP]
yvesr: does it have to be a REC?
09:17:59 [ericP]
sandro: no no
09:18:34 [ericP]
davidwood: cygri's point about how in a couple years, folks won't care makes sense. should be a sepparate doc
09:19:33 [ericP]
cygri: i can do it with another
09:20:24 [ericP]
davidwood: editors of the Concepts doc (cygri and davidwood) should edit New Features
09:20:51 [cygri]
PROSOAL: New Note: "RDF 1.1 New Features and Migration Guide"
09:20:56 [sandro]
RDF 1.1 New Features and Migration Guide
09:21:08 [davidwood]
09:21:13 [sandro]
09:21:15 [ivan]
09:21:15 [yvesr]
09:21:16 [davidwood]
09:21:17 [Arnaud]
09:21:20 [AZ]
09:21:21 [cygri]
09:21:22 [ericP]
09:21:36 [ivan]
RESOLVED: new Note: "RDF 1.1 New Features and Migration"
09:22:15 [cygri]
09:22:53 [sandro]
close issue-101
09:22:53 [trackbot]
ISSUE-101 Will RDF-WG produce a “What's New in RDF 1.1” document or migration guide? closed
09:23:58 [ericP]
ACTION: davidwood to draft “What's New in RDF 1.1” document
09:23:58 [trackbot]
Could not create new action - action title not proper UTF-8
09:24:08 [ericP]
ACTION: davidwood to draft "What's New in RDF 1.1" document
09:24:08 [trackbot]
Created ACTION-193 - Draft "What's New in RDF 1.1" document [on David Wood - due 2012-11-05].
09:25:35 [ericP]
topic: Primer
09:26:38 [ericP]
Guus: thinking about this from the persecptive of the TOC
09:27:44 [ericP]
[syntax discussion]
09:28:32 [ericP]
ivan: could use the OWL Primer trick for multiple syntaxes
09:28:50 [ericP]
Guus: should be done with the Turtle first
09:29:18 [ericP]
... RDFa and JSON-LD are good candidates. RDF/XML?
09:29:26 [ericP]
davidwood: RDF/XML will still be a REC
09:30:54 [ericP]
ACTION: Guus to add to RDF Primary and ISSUE about multi-language representation
09:30:54 [trackbot]
Created ACTION-194 - Add to RDF Primary and ISSUE about multi-language representation [on Guus Schreiber - due 2012-11-05].
09:31:19 [ericP]
cygri: TOC looks similar to the RDF Concepts' TOC
09:31:40 [ericP]
... RDF Concepts exposes the RDF data model
09:31:41 [ivan]
09:32:30 [ericP]
... should RDF Concepts have the introductory material?
09:32:50 [ericP]
... for the Primer to be as effective as possible, it should have lots of examples
09:33:16 [ericP]
... the RDF Concepts intro doesn't do that; just gives an overview of the terminology
09:34:02 [mlnt]
mlnt has joined #rdf-wg
09:34:17 [ericP]
[Guus shows the introductory text in Primer]
09:34:52 [ericP]
[general agreement that this intro is more basic than the intro in RDF Concepts]
09:35:43 [ericP]
ivan: being a visual type, i'd like to see images with discussions of graphs
09:36:00 [ericP]
Guus: will discuss embedding SVG with cygri
09:38:22 [ericP]
... aiming for minimal examples which demonstrate graphs
09:38:27 [ericP]
[general approval]
09:39:22 [Arnaud]
typo in the first sentence of section 2.1: indetified -> identified
09:41:31 [sandro]
guest: Tim Berners-Lee
09:41:32 [shh]
shh has joined #rdf-wg
09:41:44 [ericP]
davidwood: developers' eyes glaze over at the "subject, predicate, object" text.
09:41:46 [ericP]
... using the text "two things and a relationship between them" seems to work
09:42:43 [ericP]
timbl: the graph eludes people, while turtle looks like a language to them
09:43:30 [ericP]
cygri: per yvesr, would like to point at the use case.
09:44:44 [ericP]
Guus: avoiding syntax, discuss the basic concepts: iri, literal, and unfortunately the blank node
09:46:48 [yvesr]
did we just skip vocabularies?
09:47:44 [pchampin]
pchampin has joined #rdf-wg
09:48:39 [ericP]
... what can we show with basic graphs?
09:48:49 [ericP]
cygri: graph equivalence and basic entailment
09:51:28 [ericP]
cygri: if we look at datasets out there, Dublic Core is everywhere
09:53:22 [ericP]
... might be a good guide to use examples of vocabs that folks will encounter in practice
09:53:50 [ericP]
... skos is the 3rd most popular (after DC and FOAF)
09:55:11 [ericP]
Guus: i'd like to introduce in 6 or 7 printed pages
09:55:34 [ericP]
... rest in appendeces
09:59:47 [ericP]
... focusing on common vocabularies
09:59:49 [ericP]
davidwood: i find folks frequently have to invent vocabularies.
10:00:13 [yvesr]
yvesr: a way to structure the vocabulary section around use-cases would be to use DC and FOAF for people and artworks they made and create a new property using RDF Schema for 'is displayed in', to stitch two vocabularies together
10:00:13 [ericP]
... would like "of course, use others if available, but easy to invent as well"
10:00:30 [ericP]
sandro: it's like programmers using libraries
10:00:41 [ericP]
cygri: will it mention e.g. SPARQL?
10:01:16 [ericP]
ivan: there's no SPARQL primer
10:01:22 [SteveS]
SPARQL 1.1 Overview
10:01:37 [ericP]
ericP: top section of SPARQL serves as a primer
10:02:02 [sandro]
sandro: I think there should be a tiny sparql example, like the first in
10:02:11 [sandro]
... and then it links to more
10:03:07 [ericP]
Guus: remind me to point at
10:03:38 [ericP]
sandro: we should have every document published before asking for the charter extension
10:05:27 [ericP]
Guus: i need to make a statement about whether you refer to a page or a person
10:06:12 [ericP]
ivan: use hash IRIs and avoid indirection
10:06:43 [sandro]
sandro: "We use IRIs (which are basically the same as URLs) to refer to both Web Pages and things that are not Web Pages, like People and Places"
10:09:04 [ericP]
ACTION: Guus to provide Primer document for review -- due 3 Jan
10:09:04 [trackbot]
Created ACTION-195 - Provide Primer document for review -- due 3 Jan [on Guus Schreiber - due 2012-11-05].
10:09:29 [ericP]
ACTION: Guus to provide Primer document for review - due 3 Jan
10:09:29 [trackbot]
Created ACTION-196 - provide Primer document for review [on Guus Schreiber - due 2012-01-03].
10:09:38 [ericP]
drop action 4
10:09:56 [ericP]
10:10:31 [sandro]
close action-195
10:10:31 [trackbot]
ACTION-195 Provide Primer document for review -- due 3 Jan closed
10:27:57 [shh]
shh has joined #rdf-wg
10:33:09 [Arnaud]
scribe: Arnaud
10:33:26 [Arnaud]
topic: RDF/XML
10:34:19 [SteveS]
SteveS has joined #rdf-wg
10:34:49 [Arnaud]
ACTION: Fabien to provide a new WD for the WG to review by Nov 21st telecon
10:34:49 [trackbot]
Created ACTION-197 - Provide a new WD for the WG to review by Nov 21st telecon [on Fabien Gandon - due 2012-11-05].
10:36:04 [Arnaud]
davidwood: we will do the minimum
10:36:54 [davidwood]
The WG would like to vote on moving RDF/XML to FPWD on 28 Nov.
10:36:56 [Arnaud]
FabGandon: I think there are 2 issues I need to look at but agree we should keep it to a minimum, WG has a lot more important things to do
10:37:05 [ivan]
ack ivan
10:37:13 [davidwood]
10:37:37 [Arnaud]
ivan: we should aim at an edited recommendation which is easier to do
10:38:04 [Arnaud]
... allows to skip several steps
10:38:28 [Arnaud]
cygri: need to understand what it means for existing implementations
10:38:44 [Arnaud]
ivan: don't see any change in the concepts that would impact RDF/XML
10:39:29 [paul-huawei]
paul-huawei has joined #rdf-wg
10:39:41 [Arnaud]
cygri: syntax hasn't changed but the change on literals may have an impact
10:40:10 [paul-huawei]
paul-huawei has left #rdf-wg
10:40:17 [Arnaud]
... this requires more thoughts
10:40:44 [Arnaud]
ivan: yes, you're right. the section on mapping needs to change and this won't be just editorial
10:41:00 [Arnaud]
... so we can't go the EPR route
10:41:22 [davidwood]
10:41:27 [davidwood]
10:42:04 [sandro]
10:42:47 [Arnaud]
sandro: if we go to CR we'll need people to implement it, is it going to be hard to get this?
10:43:17 [Arnaud]
ivan: rdflib has already implemented it
10:43:59 [Arnaud]
davidwood: we should decide whether rdf/xml will be extended to support html5 datatype
10:44:32 [sandro]
PROPOSED: We we update RDF/XML to include special syntactic support for our new HTML datatype
10:44:38 [sandro]
10:44:46 [Arnaud]
ivan: except it's impossible to do
10:45:06 [Arnaud]
cygri: we could do it for a subset: xhtml
10:45:13 [AZ]
10:45:27 [Arnaud]
... don't see much benefits in doing this
10:45:27 [sandro]
cygri: It would only work for XHTML, and for that you can use XMLLiteral
10:45:30 [Arnaud]
ivan: I agree
10:46:18 [sandro]
+1 include examples of HTML using CDATA
10:46:18 [Arnaud]
sandro: should we have something editorial explaining how to use cdata for this?
10:46:34 [Arnaud]
ivan: probably a good idea
10:47:01 [Arnaud]
davidwood: do we want to have a paragraph on this in rdf/xml?
10:47:07 [Arnaud]
ivan: leave it to the editor
10:47:37 [Arnaud]
fabgandon: will add it to the list and present it to the WG for review
10:47:58 [Arnaud]
topic: RDF 1.1 Primer
10:48:30 [Arnaud]
guus: talking about datatypes, need to put some examples
10:48:47 [Arnaud]
... string, decimal, integer, boolean, datetime
10:48:56 [ivan]
10:48:56 [shh]
shh has joined #rdf-wg
10:48:57 [Arnaud]
... 5 key ones people typically use
10:49:05 [Arnaud]
... is that a good approach?
10:49:14 [ivan]
10:49:17 [Arnaud]
sandro: would add double
10:49:59 [ivan]
10:50:04 [Arnaud]
davidwood: ldp has a different list
10:50:49 [Arnaud]
sandro: ldp's list is based on what programing languages use
10:51:39 [Arnaud]
cygri: warn against drafting a list that gives the impression of being a recommended list, implying the rest is not recommended
10:51:43 [Arnaud]
guus: agree
10:52:21 [Arnaud]
ivan: wouldn't mention xsd:string, and keep it to plain literals
10:52:53 [Arnaud]
guus: could have a background note saying you can skip this
10:53:40 [Arnaud]
davidwood: wouldn't do more than this
10:55:33 [sandro]
18:30 doors open
10:56:52 [Arnaud]
topic: rdf 1.1 schema
10:57:12 [Arnaud]
davidwood: unlikely Dan will edit this spec, we need a new editor
10:58:37 [Arnaud]
arnaud: would be happy to help but have no background on what needs to be done so not sure I'm qualified
10:59:23 [davidwood]
11:00:30 [davidwood]
RESOLVED to mark rdf:Alt and rdf:Bag as archaic using the language at:
11:00:30 [davidwood]
11:00:46 [Arnaud]
11:00:46 [trackbot]
ISSUE-77 -- Should we mark rdf:Seq as archaic (cf ISSUE-24) -- open
11:00:46 [trackbot]
11:01:12 [Arnaud]
ivan: don't think there is a major work that needs to be done
11:02:00 [davidwood]
PROPOSAL: Close ISSUE-77 taking no action (due to XMP and RSS 1.0)
11:02:42 [ivan]
11:03:04 [Arnaud]
cygri: would like us to try to do better
11:03:43 [Arnaud]
won't we have a formal objection from Adobe?
11:04:10 [sandro]
sandro: If they were to make a new version of XMP that did not to be backward compatible, THEN they should not use ______
11:04:42 [Arnaud]
davidwood: not sure they care, they haven't touched XMP in years
11:04:57 [Arnaud]
ivan: it's not just adobe, a lot of other people use it
11:05:28 [Arnaud]
sandro: steve harris still thinks there is nothing better than seq
11:06:09 [sandro]
eric: Having two broken solutions is worse than having one broken one
11:06:28 [Arnaud]
ivan: don't think we can decide on making it archaic
11:07:12 [Arnaud]
ivan: we should provide guidance in the primer
11:07:19 [Arnaud]
... that's the place to do that
11:07:24 [davidwood]
PROPOSAL: Close ISSUE-77 taking no action (due to XMP and RSS 1.0). Instead, add guidance on avoiding Seq in the Primer.
11:07:31 [Arnaud]
sandro: would like to have it in the spec
11:09:06 [Arnaud]
cygri: tempted to say that we should talk about this is in concepts
11:09:32 [Arnaud]
... but that would be odd because it currently doesn't talk about vocabulary at all
11:10:35 [sandro]
11:10:35 [Arnaud]
ivan: maybe the schema spec is the right place
11:11:59 [Arnaud]
cygri: we should have an issue for well formed lists
11:12:17 [sandro]
my "well-formed list" proposal:
11:12:21 [Arnaud]
davidwood: so the options are: primer, schema, or possibly concepts?
11:12:39 [sandro]
(which Steve Harris didnt like)
11:13:04 [Arnaud]
... don't see a way to put in concepts without a syntax which is ugly
11:13:14 [davidwood]
Not in Concepts
11:13:28 [Arnaud]
cygri: yes, agree we should have somewhere but not in my spec! :-)
11:14:29 [Arnaud]
ivan: my feeling is that schema is probably the best place
11:14:59 [Guus]
q+ to ask about status of term "property" in Schema, Concepts, and Primer
11:15:08 [ivan]
ack ivan
11:15:17 [davidwood]
ack sandro
11:15:19 [sandro]
11:15:30 [davidwood]
ack Guus
11:15:30 [Zakim]
Guus, you wanted to ask about status of term "property" in Schema, Concepts, and Primer
11:16:46 [Arnaud]
guus: have trouble writing data model in primer because predicate is talked about in different places
11:17:01 [Arnaud]
... property vs predicate
11:17:26 [Arnaud]
... is it useful to make the distinction in primer? it's very important for schema
11:18:32 [Arnaud]
cygri: schema is a language for definiing and describing properties and classes
11:18:43 [shh]
shh has joined #rdf-wg
11:18:49 [Arnaud]
... even if rdf schema didn't exist you would still have properties
11:19:18 [Arnaud]
sandro: if we could do it all over again, it would be easier for user to call the middle slot property name
11:19:52 [yvesr]
is there an actual entailment that says :s :p :o => :p a rdf:Property?
11:19:52 [Arnaud]
cygri: agree
11:20:19 [Arnaud]
guus: alternative suggestion is to just drop the term predicate
11:21:20 [Arnaud]
cygri: you could only talk about predicates later in the document
11:21:47 [ivan]
ivan has left #rdf-wg
11:21:47 [Arnaud]
... just not use the terms subject predicate object
11:23:28 [sandro]
Issue: Shall we highlight Turtle's list structures as "Well-Formed Lists" in one of our Recs?
11:23:28 [trackbot]
Created ISSUE-102 - Shall we highlight Turtle's list structures as "Well-Formed Lists" in one of our Recs? ; please complete additional details at .
11:24:00 [cygri]
11:24:00 [trackbot]
ISSUE-77 -- Should we mark rdf:Seq as archaic (cf ISSUE-24) -- open
11:24:00 [trackbot]
11:24:14 [ivan]
ivan has joined #rdf-wg
11:24:54 [Arnaud]
sandro: would like to propose we agree to pick one single preferred solution for future developments
11:25:11 [Arnaud]
davidwood: steve would say seq
11:26:09 [Arnaud]
davidwood: don't think that works, we have different communities that like either
11:26:32 [Arnaud]
ivan: sparql could have done it but didn't
11:27:01 [Arnaud]
davidwood: ldp trying to work around this by defining a sort predicate
11:27:21 [Arnaud]
... this suggests that ordering is important
11:28:11 [Arnaud]
eric: this is hard to solve and that's why sparql didn't do it
11:29:51 [Arnaud]
ivan: there is no clear choice between the two
11:30:19 [Arnaud]
... ldp's solution is not rdf core
11:30:29 [Arnaud]
sandro: we could introduce one
11:30:40 [Arnaud]
cygri: not a good idea at this point in the process
11:30:57 [davidwood]
PROPOSAL: Close ISSUE-77 taking no action.
11:31:08 [AZ]
11:31:13 [ivan]
11:31:17 [sandro]
11:31:18 [yvesr]
11:31:37 [ericP]
11:31:42 [davidwood]
11:31:45 [Arnaud]
11:31:54 [FabGandon]
11:31:54 [Guus]
11:31:56 [cygri]
11:32:15 [Arnaud]
sandro: this is one of the hardest point in rdf
11:32:31 [Arnaud]
... this is the best opportunity to steer the community
11:33:01 [Arnaud]
yvesr: what's your preference?
11:33:08 [Arnaud]
sandro: slight preference for list
11:33:15 [Arnaud]
davidwood: I use seq!
11:33:32 [Arnaud]
11:34:21 [sandro]
11:34:35 [Arnaud]
... quite happy with sandro's proposal to say xmp is fine the way it is but if you were to do another version there is a better way
11:34:50 [Arnaud]
ivan: don't think we'll get consensus
11:35:09 [Arnaud]
sandro: steve didn't object
11:35:40 [Arnaud]
sandro: the best thing for the community is to specify one
11:36:14 [Arnaud]
... could live with anything, saying to use something else than seq would be a good start
11:36:38 [sandro]
PROPOSED: Close issue-77 marking SEQ as Archaic (saying folks shouldn't use it for new vocabs)
11:36:42 [Arnaud]
proposal: tentatively mark seq as archaic and see whether steve objects
11:36:53 [Arnaud]
ivan: ok but need to ask steve
11:36:53 [Guus]
q+ to say I was planning to use an example of the () collction notation of Turtle in the Primer
11:37:02 [davidwood]
ack Guus
11:37:02 [Zakim]
Guus, you wanted to say I was planning to use an example of the () collction notation of Turtle in the Primer
11:37:09 [cygri]
11:37:13 [sandro]
11:37:31 [AZ]
11:37:47 [davidwood]
+1 (and will take an action to ask Steve H and the Adobe AC rep)
11:37:51 [ivan]
+1 provided we cross-check with possible oppenents
11:37:53 [yvesr]
11:37:58 [Arnaud]
11:39:26 [sandro]
RESOLVED: Close issue-77 marking SEQ as Archaic (saying folks shouldn't use it for new vocabs)
11:39:27 [davidwood]
RESOLVED: Close issue-77 marking SEQ as Archaic (saying folks shouldn't use it for new vocabs)
11:40:47 [sandro]
close issue-77
11:40:47 [trackbot]
ISSUE-77 Should we mark rdf:Seq as archaic (cf ISSUE-24) closed
11:40:53 [Arnaud]
ACTION: Arnaud to check with Dan what he wants to do with regard to editorship (remain editor, leave it to Arnaud, remain co-editor?)
11:40:53 [trackbot]
Created ACTION-198 - Check with Dan what he wants to do with regard to editorship (remain editor, leave it to Arnaud, remain co-editor?) [on Arnaud Le Hors - due 2012-11-05].
11:41:09 [Arnaud]
break for lunch
11:41:21 [Arnaud]
come back at 1:30pm
11:41:29 [FabGandon]
FabGandon has left #rdf-wg
11:54:47 [ScottB]
ScottB has joined #rdf-wg
11:57:24 [Zakim]
12:01:51 [ScottB]
Zakim, Tony is temporarily me
12:01:51 [Zakim]
+ScottB; got it
12:39:53 [gavinc]
gavinc has joined #rdf-wg
12:40:55 [SteveS]
SteveS has joined #rdf-wg
12:43:26 [sandro]
Zakim, who is on the call?
12:43:26 [Zakim]
On the phone I see Rhone_4, ScottB
12:44:49 [sandro]
scribe: sandro
12:44:53 [sandro]
topic: RDF Concepts
12:45:00 [sandro]
RRSAgent, pointer?
12:45:00 [RRSAgent]
12:45:48 [sandro]
cygri: we did the bulk of the work quite some time ago. the document is in pretty good shape, with some open issues marked.
12:46:04 [sandro]
... biggest was RDF Datasets, which has become easier because we scaled down our goals.
12:46:26 [sandro]
... most of the rest is editorial. explanations. whether text should be in one spec or another.
12:46:32 [sandro]
... going throught the issues
12:46:42 [sandro]
... Defn of RDF Datasets?
12:46:59 [sandro]
... HashURIs / trig -- how do fragments work?
12:48:11 [sandro]
.... High Level Issues: Datasets, Fragments in multigraph syntax, Relationship between Semantics and Concepts documnets, and stuff in Primer vs Concepts.
12:48:35 [sandro]
subtopic: Relationship between Concepts and Primer
12:48:42 [sandro]
cygri: Section 1.
12:50:19 [sandro]
... gives a background what the datamodel is useful for, rather than just starting with concepts
12:50:25 [sandro]
... and gives context.
12:51:01 [sandro]
... There's a lot of additional terminology, used all the time in RDF, but is nowhere defined. "Resource". "denote". ...
12:51:11 [sandro]
... namespace, namespace IRI, etc
12:51:28 [sandro]
... this is all "informative".
12:51:50 [sandro]
... Does *not* give examples. That would make it bigger, and something different.
12:52:17 [sandro]
... Is it okay that intro tackles these things?
12:52:42 [Guus]
12:52:46 [ivan]
12:53:17 [sandro]
sandro: I think it's fine to have example to help the sophisticated reader understand this stuff
12:53:32 [sandro]
davidwood: what about syntax?
12:54:21 [Guus]
ack ivan
12:54:24 [sandro]
sandro: I think it's fine to use Turtle. Assume on the first reading they wont know/care about what Turtle means formally.
12:54:26 [Arnaud]
12:54:35 [sandro]
davidwood: And they'll have read Primer first.
12:54:46 [sandro]
ivan: I agree the Intro should be part of the document, yes.
12:55:01 [sandro]
ivan: Examples -- I don't know. Probably yes,
12:55:02 [Guus]
q+ to say: yes
12:55:21 [sandro]
ivan: But yes, keep Intro
12:55:29 [Guus]
ack Arnaud
12:55:37 [sandro]
Arnaud: I don't know how far we can go with graphic represetnation in examples.
12:56:07 [sandro]
.. if you can stick to diagrams, that's great
12:56:21 [sandro]
davidwood: namespace IRI
12:57:00 [sandro]
cygri: Examples should cover the stuff that's hard to visualize or grasp. That would help.
12:57:19 [sandro]
.. I'm a little bit afraid that...
12:57:40 [sandro]
.. we should make clear the difference between Primer and Concepts Introduction.
12:58:08 [sandro]
.. There's a danger if this gets to helpful, with diagrams, that people will skip the Primer.
12:58:33 [Guus]
12:58:50 [sandro]
ivan: it's a judgement call. the RDFa spec has something of a tutorial style in it, with lots of examples. And we also have a primer. A lot of people still use the primer.
12:58:54 [sandro]
12:59:08 [Arnaud]
ack Guus
12:59:08 [Zakim]
Guus, you wanted to say: yes
12:59:26 [sandro]
Guus: This section is short, so I don't see a problem. You give some definitions/references, like IRIs, that belong here.
12:59:49 [Guus]
ack Guus
12:59:56 [sandro]
... I can see your concern about examples; I think the idea of diagrams.
13:01:32 [sandro]
davidwood: Except for 1.4 (namespace IRIs) it seems like I have to read the primer first. 1.4 should stay in Concepts in order to make Concepts able to stand on its own. (Primer isn't normative.)
13:01:32 [Guus]
13:01:49 [ivan]
ack Guus
13:02:12 [sandro]
guus: The primer will be talking about Properties, so it's rather different.
13:03:46 [sandro]
13:04:17 [sandro]
"This section should be removed from RDF Concepts and folded into [RDF-SCHEMA] which actually defines the terms in question. This is ACTION-121."
13:04:48 [sandro]
cygri: All terms should be in the RDFS document, including the rdf: namespace.
13:05:12 [sandro]
.. what goes in which is entirely a historical accident.
13:05:49 [sandro]
ivan: Don't forget namespace document
13:05:57 [sandro]
sandro: indeed, my action item
13:06:57 [sandro]
cyg: Relationship to Semantics....? it'd be nice to have a principle about which text goes where.
13:07:07 [sandro]
subtopic: hash uris in datatype syntaxes
13:07:14 [sandro]
cyg: media type registration
13:07:52 [sandro]
.. all the RDF syntaxes (should) refer to section 6 in concepts which talks about how hash URIs work here.
13:07:59 [sandro]
.. does this apply in TriG as well?
13:08:13 [sandro]
.. it's "mostly" the same as in RDF.
13:08:25 [sandro]
.. but you MIGHT use it to name a graph, which is kind of a part of the document.
13:08:55 [Guus]
13:08:55 [Guus]
13:08:57 [Guus]
13:09:06 [yvesr]
sandro: we still need the graph URIs to be spelled out in the same way as other URIs
13:09:52 [sandro]
cyg: If you used fragments like file1#graph1 then that mirrors other fragment syntaxes
13:10:03 [yvesr]
13:10:04 [sandro]
sandro: but you can use them repeatedly -- that's not like in HTML or XML
13:10:41 [sandro]
cygri: to me the graph name situation, it's not external to the document, so it seems different
13:11:16 [sandro]
ivan: pragmatically speaking, this is in the RFC's to direct general processors what to do with this content. and this bit in the RFC wouldn't help
13:11:39 [pchampin]
q+ to make a crazy proposal
13:12:01 [sandro]
cygri: Web crawling use case. Trig File in the wild. Claims "heres the content of this graph" so you dont want to put it in your store. But in the fragment case, it's safer.
13:12:12 [sandro]
cygri: So there's a security thing here.
13:12:36 [Guus]
13:12:51 [sandro]
cygri: The security issue only comes up if the graph name is external to the file.
13:13:31 [Guus]
ack pchampin
13:13:31 [Zakim]
pchampin, you wanted to make a crazy proposal
13:13:59 [sandro]
pchampin: I'm guessing it's possible for a mime type to specify different meaning depending on their syntax. We could have a special syntax for fragments identifying graphs...
13:14:21 [sandro]
13:14:56 [yvesr]
13:15:00 [sandro]
pchampin: If TriG had a special syntax for identifying *graphs* in the documents, we would have a way to formally declare those URIs denote graphs.
13:15:16 [sandro]
sandro: yes it would work, but not it's not a good idea. :-)
13:15:44 [sandro]
davidwood: We've had several discussions about how we can't rely on mime types anyway.
13:15:59 [MacTed]
MacTed has joined #rdf-wg
13:17:25 [sandro]
cyg: I hear some skepticism about whether it's necessary to say anything special about fragments in dataset...
13:17:39 [davidwood]
Only 5 of the 10 issues listed at the top of Section 6 are still open
13:17:43 [yvesr]
13:18:00 [sandro]
sandro: I'd say don't try to hard, but go ahead if you like and we'll probably be fine with it.
13:18:05 [sandro]
cyg: okay.
13:18:12 [sandro]
subtopic: Abstract syntax for datasets
13:18:30 [sandro]
cygri: Section 4, pasted in SPARQL text, and issue box.
13:19:56 [sandro] 6. Abstract Syntax for Working with Multiple Graphs
13:20:18 [sandro]
cygri: maybe we can delete the issue box and leave in that definition?
13:20:41 [sandro]
davidwood: Probably soon, once issues are closed.
13:20:41 [sandro]
13:21:27 [sandro]
cygri: What we don't have in there are "gbox" and "graph store". We don't have David's circle diagram.
13:21:44 [sandro]
ivan: I think it would be useful to put here.
13:22:05 [sandro]
cyg: Not here. We don't define those things. That's more Infroamtive Intro
13:22:26 [sandro]
.. Think of mutable gbox or graph store thing, then static snapshot is an RDF Dataset.
13:23:05 [sandro]
.. "gbox is a resource that can have a snapshot that is an RDF Graph" which doesn't seem useful to say Normatively. So I'd put it in 1.5
13:23:09 [sandro]
davidwood: Agreed
13:23:23 [sandro]
cygri: I propose to just call it "gbox".
13:23:28 [sandro]
13:23:58 [sandro]
davidwood: That's not the first time. :-)
13:24:00 [sandro]
13:24:01 [sandro]
13:24:04 [sandro]
13:24:12 [sandro]
13:24:33 [sandro]
13:24:36 [trackbot]
ISSUE-5 -- Should we define Graph Literal datatypes? -- closed
13:24:36 [trackbot]
13:24:45 [sandro]
13:26:46 [sandro]
PROPOSED: Close issue-14, based on previous resolutions. We're sticking with SPARQL's notions.
13:26:47 [FabGandon]
FabGandon has joined #rdf-wg
13:27:22 [pchampin2]
pchampin2 has joined #rdf-wg
13:27:26 [cygri]
13:28:08 [yvesr]
13:29:35 [davidwood]
13:29:42 [davidwood]
davidwood has joined #rdf-wg
13:30:02 [sandro]
sandro: I have a preference for calling the pair a name-pair and calling the graph the named graph.
13:30:23 [Guus]
13:30:25 [davidwood]
+1 to Sandro
13:30:50 [sandro]
cygri: I think this is as pat says a harmless abuse of terminaology. I'm okay with some non-normative explanation of how things are used.
13:31:14 [sandro]
guus: For the primer I wouldnt get into that.
13:31:30 [sandro]
guus: Pat's think about name isn't a name.
13:31:40 [sandro]
guus: I like "labeled graph" better but it wouldn't stick.
13:33:10 [cygri]
PROPOSAL: Close ISSUE-14. We're sticking with the SPARQL definition of Named Graphs. Informative text will explain that “named” isn't to be taken too literally, and that “named graph” often refers only to the graph part of the pair.
13:33:46 [sandro]
13:33:57 [pchampin]
13:33:57 [ivan]
13:34:03 [Guus]
13:34:24 [AZ]
13:34:27 [yvesr]
+1 but would like 'not be to taken too literally' to be rephrased
13:34:38 [yvesr]
s/be to/to be/
13:34:41 [sandro]
cyg: "despite the use of the word 'name', the IRI does not necessarily denote the graph" (or something like that)
13:35:17 [yvesr]
13:35:23 [Arnaud]
13:35:26 [davidwood]
+1 given Richard's explanation
13:35:33 [FabGandon]
13:36:15 [sandro]
close issue-14
13:36:20 [davidwood]
RESOLVED: Close ISSUE-14. We're sticking with the SPARQL definition of Named Graphs. Informative text will explain that “named” isn't to be taken too literally, and that “named graph” often refers only to the graph part of the pair.
13:36:31 [cygri]
13:36:49 [sandro]
13:36:55 [davidwood]
ISSUE-14 closed via the Web
13:37:04 [sandro]
PROPOSED: Close issue-15, the relationship is undefined.
13:37:10 [ivan]
13:37:11 [cygri]
13:37:12 [AZ]
13:37:21 [davidwood]
13:37:27 [sandro]
13:37:27 [cygri]
Noting that this follows from the earlier resolution of not defining dataset semantics
13:37:42 [Arnaud]
13:37:50 [davidwood]
RESOLVED: Close issue-15, the relationship is undefined.
13:37:53 [cygri]
13:38:53 [sandro]
cyg: RDF Semantics says how to merge RDF Graphs. What about RDF Datasets?
13:39:16 [davidwood]
ISSUE-15 closed via the Web
13:40:41 [sandro]
PROPOSED: close issue-17 -- there is no general purpose way to merge datasets; it can only be done with external knowledge.
13:40:57 [sandro]
yvesr: Do we need dataset isomorphism?
13:41:12 [sandro]
cyg: It's in there for graphs for testing, beside that do we need it.
13:41:33 [sandro]
cyg: We might need something for testing, later on, yes.
13:41:46 [cygri]
13:41:47 [sandro]
13:41:47 [ivan]
13:41:48 [AZ]
13:41:51 [FabGandon]
13:41:53 [Guus]
13:41:56 [yvesr]
+1 but we should spell it out
13:42:01 [Arnaud]
13:42:03 [davidwood]
13:42:45 [sandro]
RESOLVED: close issue-17 -- there is no general purpose way to merge datasets; it can only be done with external knowledge.
13:43:16 [davidwood]
Closed ISSUE-17 via the Web
13:43:37 [sandro]
yvesr: How about in a note?
13:43:51 [sandro]
ivan: yes, in some Dataset Semantics note, this stuff appears
13:43:53 [cygri]
13:44:02 [sandro]
sandro: yes, it falls out pretty much automatically.
13:44:09 [Zakim]
13:45:02 [davidwood]
13:46:06 [sandro]
sandro: I thought we resolved TriG would { } empty graphs
13:46:13 [path]
path has joined #rdf-wg
13:46:19 [sandro]
cygri: SPARQL allows empty graphs.
13:46:44 [davidwood]
13:46:45 [sandro]
cygri: I hope at least the abstract syntax has empty graphs.
13:47:01 [davidwood]
RESOLVED: Our dataset syntax will allow for the expression of empty named graphs, whatever their semantics might be.
13:47:06 [sandro]
(from before)
13:47:33 [sandro]
yvesr: The assymmetry between N-Quads and TriG
13:47:45 [path]
And their semantics is, they are trivially true in all interpretations.
13:47:55 [sandro]
cygri: SPARQL handles that by saying stores can treat an empty graph as not existing.
13:48:33 [davidwood]
Closed ISSUE-22 based on the previous resolution
13:49:06 [sandro]
cygri: "Note: N-Quads can't record the existings of empty names graphs" ... "Tools may not want to attach too much meaning to empty graphs"
13:49:36 [sandro]
sandro: empty default graph?
13:49:54 [path]
But Nquads can do this. One can use nulls, for example.
13:49:59 [Zakim]
13:50:02 [sandro]
cyg: THeree is always a default graph
13:50:06 [Zakim]
13:50:20 [sandro]
we do not hear Gavin
13:50:31 [yvesr]
cygri: this is a TriG issue so should be tackled ther
13:50:40 [yvesr]
13:51:05 [sandro]
guus: People who use this might expect information loss
13:51:23 [path]
I want to see the video of gavin merging with himself
13:52:48 [cygri_]
cygri_ has joined #rdf-wg
13:52:53 [Zakim]
13:53:11 [trackbot]
trackbot has joined #rdf-wg
13:53:17 [sandro]
trackbot, issue-1?
13:53:20 [trackbot]
Sorry, sandro, I don't understand 'trackbot, issue-1?'. Please refer to for help
13:53:45 [cygri_]
ACTION: cygri to add note on RDF Datasets noting that empty named graphs may not survive in some implementations
13:53:54 [trackbot]
Created ACTION-199 - Add note on RDF Datasets noting that empty named graphs may not survive in some implementations [on Richard Cyganiak - due 2012-11-05].
13:55:16 [sandro]
PROPOSED: Close-32 -- we do not provide a standard way to identify g-boxes or g-snaps, but we recognize that people can do it by defining predicates that let them do it. We may publish a NOTE which provides such a vocabulary.
13:55:43 [cygri]
13:55:46 [ivan]
13:55:46 [sandro]
13:55:47 [davidwood]
13:55:47 [Guus]
13:55:47 [yvesr]
13:55:51 [FabGandon]
13:55:55 [pchampin]
13:55:56 [AZ]
13:56:05 [davidwood]
RESOLVED: Close-32 -- we do not provide a standard way to identify g-boxes or g-snaps, but we recognize that people can do it by defining predicates that let them do it.  We may publish a NOTE which provides such a vocabulary.
13:56:15 [sandro]
close issue-32
13:56:15 [trackbot]
ISSUE-32 Can we identify both g-boxes and g-snaps? closed
13:56:18 [path]
13:56:18 [sandro]
13:56:21 [trackbot]
ISSUE-29 -- Do we support SPARQL's notion of "default graph"? -- closed
13:56:21 [trackbot]
13:57:59 [sandro]
sandro: This resolution text is just a note to ourselves to remind ourselves why we closed the resolution.
13:58:39 [sandro]
patL: I think the resolution should end at " we do not provide a standard way to identify g-boxes or g-snaps."
13:58:49 [path]
13:58:58 [AZ]
13:59:00 [sandro]
PROPOSED: Close ISSUE-32 -- We do not provide a standard way to identify g-boxes or g-snaps.
13:59:05 [ivan]
13:59:07 [AZ]
13:59:10 [Arnaud]
13:59:11 [davidwood]
PROPOSED: Close-32 -- we do not provide a standard way to identify g-boxes or g-snaps. However, we recognize that people can identify graphs in a dataset syntax by defining predicates that let them do it.  We may publish a NOTE which provides such a vocabulary.
13:59:11 [yvesr]
13:59:12 [FabGandon]
13:59:12 [sandro]
+0.5 see IRC discussion here.
13:59:13 [Guus]
13:59:51 [sandro]
sandro: Pat, are you okay with David's phrasing?
14:00:42 [sandro]
pat: Here's my problem. I am still under the impression that we have not come to any agreement as a WG. We have not chosen a vocab for this -- Sandro suggested one -- but we couldn't agree on it.
14:00:45 [davidwood]
PROPOSED: Close ISSUE-32 -- We do not provide a standard way to identify g-boxes or g-snaps.
14:00:45 [AZ]
+1 to the short version
14:00:45 [sandro]
PROPOSED: Close ISSUE-32 -- We do not provide a standard way to identify g-boxes or g-snaps.
14:00:52 [sandro]
14:00:53 [davidwood]
14:00:55 [AZ]
14:00:55 [FabGandon]
14:00:57 [ivan]
14:00:57 [yvesr]
14:00:59 [path]
14:01:00 [Arnaud]
14:01:03 [pchampin]
14:01:06 [cygri]
14:01:08 [Guus]
14:01:14 [Guus]
oops typo
14:01:15 [sandro]
ivan: No one does anything but vote until we say resolved
14:01:18 [Guus]
14:01:21 [sandro]
sandro: like scribe?
14:01:26 [gavinc]
14:01:37 [sandro]
RESOLVED: Close ISSUE-32 -- We do not provide a standard way to identify g-boxes or g-snaps.
14:02:46 [gavinc]
Yeah, we closed them by not having interoperability! Well done us! ;)
14:02:47 [path]
THis could be a model for quite a number of thorny problems.
14:02:48 [Zakim]
+ +1.617.838.aaaa
14:02:58 [MacTed]
aaaa is me
14:03:04 [shh]
shh has joined #rdf-wg
14:03:05 [MacTed]
Zakim, aaaa is me
14:03:05 [Zakim]
+MacTed; got it
14:03:15 [ivan]
zakim, who is here?
14:03:15 [Zakim]
On the phone I see Rhone_4, ScottB, PatH, GavinC.a, MacTed
14:03:16 [Zakim]
On IRC I see shh, trackbot, cygri, path, davidwood, FabGandon, MacTed, SteveS, gavinc, ScottB, ivan, pchampin, mlnt, AndyS, Guus, AZ, Arnaud, RRSAgent, Zakim, yvesr, manu1,
14:03:16 [Zakim]
... mischat, sandro, ericP
14:03:35 [ivan]
zakim, GavinC.a is really Gavinc
14:03:35 [Zakim]
+Gavinc; got it
14:03:53 [shh]
shh has left #rdf-wg
14:04:28 [sandro]
cyg: New WDs of rdf-concepts. I can address the yellow boxes now, I think.
14:04:49 [sandro]
.. Can we go to Last Call? Or put out another WD first?
14:05:05 [path]
In 2004 the semantics went through c. 30 drafts.
14:05:29 [sandro]
guus: We owe it to the community to do a WD before Last Call
14:05:39 [sandro]
sandro: why?
14:05:45 [sandro]
ivan: LC is just we're internally resolved on it, not community
14:05:54 [path]
Not a Call if you just put it out in public and invite comments.
14:06:01 [sandro]
sandro: You can go straight to last call
14:06:18 [path]
14:06:21 [davidwood]
It is a Call because we are calling for public comments
14:06:32 [sandro]
cyg: Pragmatically -- it would be bad to publish LC of Concepts since we don't even have FPWD of everything else.
14:06:39 [gavinc]
+1 to not publishing a last call for data sets without a concrete syntax for datasets
14:06:41 [sandro]
ivan: true
14:07:04 [path]
+1 to cyg.
14:07:25 [sandro]
sandro: So, next WD of rdf-concepts within a month or so?
14:07:35 [sandro]
david: (reads W3C Process on Last Call)
14:08:36 [sandro]
cyg: Yes.
14:09:00 [sandro]
topic: RDF Semantics
14:10:05 [sandro]
zakim, who is on the call?
14:10:05 [Zakim]
On the phone I see Rhone_4, ScottB, PatH, Gavinc, MacTed
14:10:14 [Zakim]
14:10:15 [sandro]
14:10:22 [gavinc]
enjoy your coffee
14:10:43 [gavinc]
7:30 PDT
14:10:52 [MacTed]
14:12:04 [Zakim]
14:12:49 [gavinc]
I have no idea how to provide any examples in TriG btw, give our resolution to ISSUE-15 :\
14:20:29 [gkellogg]
gkellogg has joined #rdf-wg
14:27:22 [Zakim]
14:27:27 [gkellogg]
zakim, I am ??P8
14:27:27 [Zakim]
+gkellogg; got it
14:27:40 [Guus]
welcome gregg, we're just in a break
14:28:45 [Guus]
we're 30 min late wrt our agenda
14:29:05 [Zakim]
14:30:19 [Zakim]
14:31:26 [Zakim]
14:32:30 [Zakim]
14:33:21 [path]
great sound tapestry of multiple languages.
14:33:28 [davidwood]
14:33:37 [gkellogg]
zakim: mute me
14:33:57 [gkellogg]
zakim, mute me
14:33:57 [Zakim]
gkellogg should now be muted
14:34:33 [cygri]
zakim, who is on the phone?
14:34:33 [Zakim]
On the phone I see Rhone_4, Gavinc, gkellogg (muted), Tony, PatH, MacTed
14:34:34 [path]
I once had a very bad curry in Marseilles.
14:34:50 [yvesr]
14:35:11 [ScottB]
Zakim, Tony is temporarily me
14:35:11 [Zakim]
+ScottB; got it
14:35:21 [MacTed]
Zakim, mute me
14:35:21 [Zakim]
MacTed should now be muted
14:35:42 [AZ]
topic: Semantics
14:35:58 [gavinc]
mmm, yes indeed as I have meetings at 9am PDT
14:36:03 [AZ]
Guus: but we'll start Turtle at 4:30 at the latest
14:36:33 [gavinc]
that means 30 minutes for Turtle yes?
14:36:39 [AZ]
Guus: we may not reach issues re. Semantics VS Concepts
14:37:04 [AZ]
... we hope to get to LC before end of Jan.
14:37:24 [AZ]
... let's look at things to be resolved in semantics
14:37:36 [AZ]
... let's have Pat's view
14:37:56 [davidwood]
GavinC, Turtle in 53 minutes
14:38:00 [davidwood]
from now
14:38:11 [pchampin]
pchampin has left #rdf-wg
14:38:39 [AZ]
path: the situation as I see it...
14:38:52 [AZ]
... there several major changes that we resolved not doing
14:39:24 [AZ]
... but regarding the doc it would be good to scale down the text
14:39:47 [Guus]
Open issues on RDF Semantics:
14:39:50 [AZ]
... now it contains both the formal and the "tutorial-like" part on semantics
14:40:12 [AZ]
... let us separate the MT in a short doc
14:40:17 [pchampin]
pchampin has joined #rdf-wg
14:40:39 [AZ]
... and do a separate doc that summarise the inference rules and give the tutorial stuff
14:41:11 [AZ]
... and declare that these rules are non normative
14:41:15 [cygri]
14:41:28 [sandro]
14:41:34 [cygri]
q+ to ask about the third rule "tutorial on model theory"
14:41:55 [sandro]
+1 splitting RDF Semantics into Normative part and Entailments.
14:41:57 [davidwood]
ack Guus
14:41:58 [AZ]
... the proof of completeness of rules etc would be removed
14:42:15 [Guus]
ack cygri
14:42:15 [Zakim]
cygri, you wanted to ask about the third rule "tutorial on model theory"
14:42:31 [AZ]
cygri: good analysis of the current docs
14:42:50 [AZ]
... I like having a more focused doc that only treats the formal stuff
14:43:52 [AZ]
... we should retain the rest in some form as a note
14:44:08 [AZ]
... for people who want to know a bit more without having to get into MT
14:44:26 [davidwood]
Good idea
14:44:29 [AZ]
... I agree the rules should not be normative
14:44:51 [AZ]
... the question is where the informative, more accessible things would go
14:45:38 [ivan]
14:46:05 [AZ]
... About the "tutorial-like" parts on model theory, I found it somehow useful in some way
14:46:31 [Guus]
14:46:38 [Guus]
ack ivan
14:46:46 [AZ]
path: we could keep the intro to MT in RDF semantics before formal definitions comes
14:47:15 [AZ]
ivan: I agree we can keep the intro on MT in RDF semantics as informative part
14:47:33 [AZ]
... and we should put references to relevant books
14:47:39 [cygri]
14:48:06 [Guus]
14:48:13 [AZ]
... the rules would end up as a note
14:48:45 [gavinc]
MT would be another document, right? Okay, yes/
14:49:39 [AZ]
... there are defintions in the MT (lean graphs, issues on bnodes, etc) which are normative so should be in RDF semantics
14:49:51 [AZ]
... but may be needed in the doc on rules
14:50:11 [Guus]
q+ to point out danger of doc proliferation
14:50:24 [AZ]
path: maybe we could keep things in one document
14:50:44 [AZ]
ivan: the different sections are for different audiences so it's better separated
14:51:05 [AZ]
Guus: I don't like document proliferation
14:51:15 [davidwood]
14:51:20 [davidwood]
ack gus
14:51:23 [AZ]
... it's a burden on our readership
14:51:24 [davidwood]
ack guus
14:51:24 [Zakim]
Guus, you wanted to point out danger of doc proliferation
14:51:30 [path]
maybe definitions of lean, merge etc,, should be in concepts?
14:51:38 [cygri]
14:51:41 [AZ]
... I feel better with Pat's idea of keeping all this in one doc
14:51:49 [Guus]
ack davidwood
14:52:12 [AZ]
davidwood: it's also a problem for our readers if the docs become insanely huge
14:52:18 [Guus]
ack cygri
14:52:51 [AZ]
cygri: it makes sense to group content of docs in function of the readership rather than in terms of theme
14:52:58 [path]
+1 to cygri. documents are aimed at readership.
14:53:05 [sandro]
+1 cygri: group documents based on readership. so keeping all three together isn't such a good idea
14:53:18 [Guus]
ok, happy to be convinced
14:53:24 [AZ]
... keeping content of RDF semantics as it is is not a good idea
14:53:44 [AZ]
... although I agree doc proliferation is also a problem
14:54:04 [MacTed]
I generally prefer three distinct 50 page docs over one 150 page doc with three distinct chapters
14:54:09 [AZ]
Guus: happy to be convinced if there are very good reasons
14:54:29 [davidwood]
q+ re RAISED issues
14:54:33 [MacTed]
(especially when they're commonly presented as one giant streaming HTML page)
14:54:41 [AZ]
... let us look at the open issues
14:54:48 [gavinc]
14:55:12 [AZ]
... open and raised issues
14:55:21 [davidwood]
14:56:10 [AZ]
path: the rules are incomplete and we should remove proofs and not express guarantees that they are complete
14:56:59 [AZ]
... there are rules that are computationally bad but necessary for completeness
14:57:25 [AZ]
cygri: could we transform this into an action
14:57:52 [AZ]
gavinc: we could make a resolution without closing the issue
14:58:37 [AZ]
Guus: we resolve to close the issue by having an action to solve it
14:59:48 [cygri]
PROPOSAL: Semantics editors accept an action to add the missing inference rule as pointed out in ISSUE-9; this closes ISSUE-9
15:00:19 [davidwood]
Opened ISSUEs 84, 85, 90, 97 and 98
15:00:27 [ivan]
15:00:29 [AZ]
15:00:30 [MacTed]
15:00:33 [davidwood]
15:00:44 [yvesr]
15:00:45 [AZ]
RESOLVED: Semantics editors accept an action to add the missing inference rule as pointed out in ISSUE-9; this closes ISSUE-9
15:00:54 [davidwood]
15:01:21 [AZ]
action: path to add the missing inference rules as pointed out in ISSUE-9; this closes ISSUE-9
15:02:18 [davidwood]
Closed ISSUE-9 via the Web with the resolution comments (lost Tracker again)
15:02:24 [trackbot]
trackbot has joined #rdf-wg
15:02:35 [AZ]
cygri: we resolved that the list of datatypes should be removed from semantics and put in concepts
15:02:40 [sandro]
action: path to add the missing inference rules as pointed out in ISSUE-9; this closes ISSUE-9
15:02:55 [AZ]
ivan: but XMLliterals is the source of inconcistency
15:03:24 [AZ]
path: we don't need to have the list of datatypes and can refer to concepts
15:04:17 [trackbot]
trackbot has joined #rdf-wg
15:04:20 [sandro]
action: path to add the missing inference rules as pointed out in ISSUE-9; this closes ISSUE-9
15:04:32 [cygri]
q+ to talk about semantic extensions
15:05:02 [AZ]
ivan: the POWDER IRI set (????) is an extension of RDF semantics
15:05:46 [AZ]
... it would be nice to have all things in one place
15:06:00 [AZ]
path: it would be a mistake to try to put everything together as it can grow
15:06:18 [AZ]
... we could say that they are legal extensions
15:06:29 [sandro]
15:06:37 [AZ]
... and specify well what are formally semantic extensions
15:06:41 [sandro]
15:07:01 [Guus]
ack cygri
15:07:01 [Zakim]
cygri, you wanted to talk about semantic extensions
15:07:34 [AZ]
cygri: the notion of semantic extension is not very well put in the doc but is mentionned
15:07:43 [AZ]
... and IRIset semantics is a very good example of extension
15:08:17 [AZ]
... rdf:plainLiteral is now irrelevant with our decision on removing plain literals
15:08:48 [Guus]
ack sandro
15:08:48 [AZ]
sandro: it's more important for OWL where you need this for property rangs
15:08:57 [AZ]
15:11:04 [sandro]
15:11:44 [AZ]
sandro: we need better explanations on what entailment is useful for the semantic web
15:12:29 [AZ]
... it seems that at the moment people have different expectations of what the semantics is for
15:13:45 [davidwood]
ack sandro
15:13:48 [ivan]
PROPOSED: the Semantics document will make the notion of semantics extension more explicit, and will refer to the POWDER and the SPARQL Entailment Regimes as good examples
15:14:19 [cygri]
15:14:23 [AZ]
15:14:25 [ivan]
15:14:26 [davidwood]
15:14:27 [Guus]
15:14:30 [FabGandon]
15:14:31 [path]
15:14:36 [yvesr]
15:14:42 [sandro]
+1 but I'm scared about what that text might look like without a lot more discussion.
15:14:51 [ivan]
RESOLVED: the Semantics document will make the notion of semantics extension more explicit, and will refer to the POWDER and the SPARQL Entailment Regimes as good examples
15:14:52 [AZ]
RESOLVED: the Semantics document will make the notion of semantics extension more explicit, and will refer to the POWDER and the SPARQL Entailment Regimes as good examples
15:16:03 [AZ]
Guus: do we postpone this for later?
15:16:17 [AZ]
ivan: rdf:plainLiteral does not change anything
15:16:38 [AZ]
... to the semantics document
15:17:31 [AZ]
... the question is only about the influence on RDF Semantics 1.1
15:18:23 [AZ]
path: any decision refering to rdf:plainLiteral would be a trivial change like 1 sentence
15:18:35 [gavinc]
Close without prejudice?
15:18:43 [AZ]
cygri: we should see if the OWL WG really care about what we do wrt plainLiteral
15:18:57 [AZ]
ivan: they probably do not care
15:19:32 [sandro]
sandro: (thinking: so, no one cares about whether rdf:PlainLiteral is "ugly"....)
15:19:53 [davidwood]
Close ISSUE-11 RESOLVED: the Semantics document will make the notion of semantics extension more explicit, and will refer to the POWDER and the SPARQL Entailment Regimes as good examples
15:20:34 [davidwood]
Close ISSUE-11 via the Web
15:21:50 [davidwood]
Closed ISSUE-70 via the Web
15:22:50 [AZ]
15:23:33 [AZ]
path: I don't know the difference between equality and identity
15:23:49 [AZ]
sandro: consider it's relation1 and relation2, and see what you get
15:24:35 [AZ]
path: let us say that RDF semantics is defined in terms of identity
15:25:20 [davidwood]
15:25:57 [AZ]
ericP: SPARQL has operations that use identity and others equality
15:26:49 [davidwood]
Alex Hall's Review of XSD Datatypes 1.1 Changes
15:27:18 [gavinc]
SPARQL has both
15:28:03 [AZ]
ericP: SPARQL has eq and =
15:28:33 [cygri]
15:28:44 [AZ]
ericP: it's the distinction you have in lisp
15:28:45 [gavinc]
1 == 1.0 or 1 != 1.0
15:29:38 [cygri]
15:29:57 [ericP]
-> SPARQL sameTerm operator
15:30:05 [AZ]
davidwood: SPARQL distnguishes "2 numbers being equal" (=) and "2 strings are equal", the later requires computation
15:31:33 [davidwood]
Almost all computer languages make the same type of comparisons: Numbers and strings are compared differently.
15:31:55 [AZ]
cygri: ask alex (Alex Hall) from SPARQL 1.1 WG to know what he thinks the clarification should be
15:33:28 [sandro]
15:34:28 [gavinc]
Yes, computers are wrong ;)
15:34:29 [sandro]
sandro: In my book this is what D-Entailment is about. Do "01"^^xs:integer and "1"^^xs:integer co-refer ?
15:34:35 [davidwood]
a = a + 1
15:35:15 [Arnaud]
this is about a == b vs a.equals(b)
15:35:24 [AZ]
cygri: there may not be interest in discussing an issue we don't even know what it is about
15:35:33 [davidwood]
+1 to Arnaud noting that neither is =
15:35:40 [cygri]
15:36:24 [cygri]
PROPOSAL: Close ISSUE-98 because we are leaving the semantics of RDF datasets unspecified
15:36:49 [davidwood]
Closed ISSUE-98
15:37:21 [gavinc]
tomorrow is much worse
15:37:24 [AZ]
Guus: we leave the other issues on semantics for later
15:37:29 [AZ]
15:37:32 [ScottB]
I'm going to have to sign off early
15:37:39 [Zakim]
15:37:54 [path]
OK, I will drop out now. thanks.
15:38:02 [Zakim]
15:38:11 [Zakim]
15:38:20 [AndyS]
zakim, IPCaller is me
15:38:20 [Zakim]
+AndyS; got it
15:38:30 [AZ]
gavinc: the remaining issues are around test cases, how we publish them
15:39:05 [AZ]
Guus: we plan to have a CR in 2 weeks, which seems unlikely
15:40:00 [AZ]
ericP: we still have Tim's comment
15:41:34 [AZ]
ivan: we can say the commenters we wait for 2 weekd for their answer and move on
15:42:12 [AZ]
ericP: if we don't get response, we assume agreement
15:43:04 [AZ]
ivan: tests don't have to be complete when we are in CR
15:43:34 [AZ]
... we have to say what we think is necessary to pass the test
15:44:18 [gavinc]
Two implementations are likely to pass the tests, in reality 3 :P gkellogg, gavinc, and AndyS :P
15:45:01 [AZ]
ivan: the call for implementation is the CR
15:45:24 [AZ]
... but we can skip it if you already have the required implementation already
15:45:34 [AZ]
sandro: but we don't have the test suite yet
15:45:40 [gavinc]
15:46:03 [AZ]
ericP: we want 2 implementations
15:46:18 [sandro]
PROPOSED: Our CR exit criterion for Turtle will be: two or more implementations passing all the approved tests in the test suite.
15:46:25 [davidwood]
15:46:27 [gkellogg]
15:46:28 [ericP]
15:46:29 [yvesr]
15:46:30 [gavinc]
15:46:33 [ivan]
15:46:33 [Arnaud]
15:46:34 [AZ]
15:46:37 [sandro]
15:46:53 [sandro]
RESOLVED: Our CR exit criterion for Turtle will be: two or more implementations passing all the approved tests in the test suite.
15:46:55 [Arnaud]
15:47:09 [Guus]
ack Arnoud
15:47:16 [davidwood]
ack Arnaud
15:47:23 [gavinc]
At risk features, PREFIX/BASE
15:47:36 [AZ]
Arnaud: do we have an idea of when we are likely to have the 2 implemetnations
15:47:46 [gkellogg]
zakim, unmute me
15:47:46 [Zakim]
gkellogg should no longer be muted
15:48:02 [AndyS]
15:48:07 [gavinc]
Do at RISK features need to be resolved before CR?
15:48:11 [AZ]
sandro: we should have a rec before asking for more time
15:48:23 [Guus]
ack AndyS
15:48:33 [gavinc]
1000 ;)
15:48:37 [gavinc]
yes it does
15:48:46 [sandro]
15:49:03 [AZ]
AndyS: we need to define formally what the test suite is
15:49:28 [AZ]
ericP: I expect it to be like the RDF/XML test suite
15:50:02 [AZ]
... it gives correspondence between RDF/XML input and N-triples equivalent
15:50:05 [gavinc]
Test Suite N-Triples
15:50:42 [AZ]
sandro: there can be negative syntax test
15:50:49 [AndyS]
+1 to Sandro
15:51:15 [AZ]
... parsers can accept invalid input but we can have the test nonetheless
15:51:49 [gkellogg]
my parser implements a "strict" option, for this purpose
15:51:54 [yvesr]
+1 to sandro as well
15:52:14 [AZ]
ericP: we could say we require 2 turtle parsers that succeed on positive test and reject negative tests
15:52:18 [gavinc]
my parser has "explode_violently"
15:52:25 [AndyS]
as gkellogg -- RIOT has special strict mode.
15:52:29 [gavinc]
steaming nature bing the issue
15:52:42 [sandro]
sandro: I think it would be good to have some "validating" parsers, parsers which can pass negative syntax tests.
15:52:55 [AZ]
ericP: negative test make sure that some things are not misinterpreted by some parsers
15:53:37 [sandro]
gavin: we need to be able to emit some triples before failing
15:53:39 [AZ]
gavinc: issues with streaming parsers
15:53:48 [AndyS]
some triples and an error is a "no"
15:53:53 [sandro]
+1 allow parser to emit triples before saying NOT-TURTLE.
15:54:01 [AZ]
ericP: I don't think streaming would be problem
15:54:25 [gkellogg]
Yes, I'll emit triples too, but raise an error on invalid input.
15:54:32 [AZ]
sandro: we are ok with having negative syntax test
15:54:34 [AndyS]
(or read to tmp place)
15:54:37 [AZ]
15:55:06 [sandro]
consensus -- we're okay with having negative syntax tests, as long as it's okay to emit some triples before rejecting the input as not strictly turtle
15:55:35 [AZ]
ericP: catalogue the tests we have in terms of features tested
15:56:29 [AZ]
... test points to relevantdocument parts
15:56:45 [AZ]
s/relevantdocument/relevant document/
15:57:51 [gkellogg]
Serialize "foo" or "foo"^^xsd:string?
15:58:18 [sandro]
subtopic: can a turtle parser normalize xsd data?
15:58:27 [AZ]
ericP: if we write the tests according to their equivalent in N-triples, the Turtle will look very much like the N-triples
15:58:44 [sandro]
eric: we could just make sure none of the test push against this.
15:59:01 [sandro]
eric: So, let's just test canonical forms for datatype literals.
15:59:12 [sandro]
+1 that should work
16:00:01 [sandro]
eric: XSD 1.0 vs XSD 1.1 doubles with trailing zero
16:00:15 [AndyS]
write the tests so NT has the exact same lexical form as input.
16:00:48 [sandro]
sandro: Let's just avoid anything like "1."^^xs:double or "1.0"^^xs:double
16:02:02 [AZ]
Guus: who's going to define the tests and the features tested
16:02:47 [AZ]
gavinc: I'd be happy to do that but it would be good to have support from someone else
16:02:55 [AZ]
gkellogg: ok to do it
16:02:58 [sandro]
+1 gkellogg doing turtle test results reporting!
16:03:12 [AndyS]
16:03:23 [gavinc]
16:04:28 [sandro]
+1 yes, the report generation -- making and colating EARL reports
16:04:59 [gavinc]
can help with hosting as well
16:05:25 [AZ]
sandro: implementers should send EARL report or at least the tests passed
16:05:30 [AndyS]
16:05:48 [sandro]
sandro: I'm just asking for someone to be accepting the list of tests pasts, and generating a page of red/green test results.
16:06:35 [ericP]
ACTION: ericP to document the testable features of Turtle
16:07:11 [AZ]
Guus: we don't need to resolve featurse at risks
16:07:12 [sandro]
Guus: gkellogg will maintain the implementation/test-results report
16:07:26 [AZ]
16:07:31 [AndyS]
err .. "at risk" is OK for LC ... but CR?
16:09:02 [AZ]
sandro: at risk means we wait for implementations to see
16:10:14 [gavinc]
16:10:28 [AndyS]
I thought NT was to be a separate doc.
16:11:11 [gavinc]
16:12:44 [sandro]
16:12:44 [gavinc] feature at risk
16:12:44 [cygri]
cygri has joined #rdf-wg
16:13:38 [gavinc]
-0.9 :P so no you can't resolve it now :P
16:13:50 [cygri]
scribe: cygri
16:14:26 [cygri]
[discussion of tests]
16:14:37 [cygri]
16:14:39 [Guus]
16:15:00 [cygri]
Arnaud: Turtle's conformance section forces us to do negative tests
16:15:13 [gavinc]
"This specification does not define how Turtle parsers handle non-conforming input documents."
16:15:15 [gavinc]
No, it doesn't.
16:15:25 [sandro]
16:15:25 [ericP]
16:15:28 [cygri]
sandro: are you looking at the right version?
16:15:43 [cygri]
Arnaud: no, sorry, nevermind
16:16:21 [Zakim]
16:17:03 [cygri]
topic: Admin
16:17:17 [Zakim]
16:17:19 [cygri]
ivan: I'm looking at the open issues for the WG
16:17:33 [cygri]
... there are more open graphs issues
16:17:49 [cygri]
16:17:51 [sandro]
16:17:53 [cygri]
16:17:55 [cygri]
16:17:58 [cygri]
16:18:09 [cygri]
16:18:16 [sandro]
23 == Does going from single-graph to multi-graph require new format and new media types?
16:18:17 [Zakim]
16:18:56 [cygri]
guus: do we have discussed concepts-semantics-relationship sufficiently?
16:18:58 [cygri]
cygri: I think yes
16:19:08 [trackbot]
trackbot has joined #rdf-wg
16:19:13 [cygri]
16:19:13 [trackbot]
ISSUE-23 -- Does going from single-graph to multi-graph require new format and new media types? -- open
16:19:13 [trackbot]
16:19:36 [cygri]
16:20:19 [cygri]
ack me
16:20:34 [cygri]
cygri: JSON-LD is currently a single- and multigraph syntax
16:20:40 [cygri]
guus: ok, let's not discuss now
16:20:42 [sandro]
cygri: JSON-LD is multigraph and single graph!
16:20:42 [sandro]
sandro: damn.
16:20:46 [cygri]
16:20:46 [trackbot]
ISSUE-31 -- Do we produce a standard (REC) syntax for conveying multiple graphs? -- open
16:20:46 [trackbot]
16:21:09 [cygri]
ivan: this is to be closed, we have a resolution
16:21:22 [cygri]
16:21:22 [trackbot]
ISSUE-35 -- Should there be an rdf:Graph construct, or something like that? -- open
16:21:22 [trackbot]
16:21:40 [cygri]
sandro: we decided to do that maybe in a Note
16:21:45 [cygri]
... so it can be closed
16:21:47 [cygri]
16:21:50 [cygri]
16:21:50 [trackbot]
ISSUE-38 -- What new vocabulary should be added to RDF to talk about graphs? -- open
16:21:50 [trackbot]
16:21:58 [sandro]
issue-31 is admin -- we've already agreed on stuff that closes this.
16:22:48 [sandro]
PROPOSED: close ISSUE-35 and ISSUE-28 saying we're not going to do these in a REC, but might something like this in a WG NOTE
16:22:58 [sandro]
PROPOSED: close ISSUE-35 and ISSUE-38 saying we're not going to do these in a REC, but might something like this in a WG NOTE
16:23:05 [yvesr]
16:23:11 [sandro]
16:23:12 [Guus]
16:23:13 [ivan]
16:23:14 [cygri]
16:23:15 [pchampin]
16:23:29 [yvesr]
(but noting it would be great to have that note :) )
16:23:32 [ericP]
16:23:34 [cygri]
RESOLVED: close ISSUE-35 and ISSUE-38 saying we're not going to do these in a REC, but might something like this in a WG NOTE
16:23:50 [cygri]
16:24:41 [cygri]
16:24:41 [trackbot]
ISSUE-95 -- Turtle Inverse Property Syntax -- open
16:24:41 [trackbot]
16:26:01 [davidwood]
Closed ISSUE-31 via the Web
16:26:12 [cygri]
cygri: we have issues left in RDF General and Cleanup tasks
16:26:12 [davidwood]
The resolution was here: RESOLVED: We will produce a W3C Recommendation for a dataset syntax, similar to TriG and to SPARQL's named graph syntax. This does not preclude recommending a syntax like n-quads.
16:26:23 [cygri]
guus: we should go through all remaining open issues tomorrow
16:27:00 [cygri]
davidwood: we have mostly syntax issues left
16:27:31 [davidwood]
…and Notes
16:28:47 [cygri]
[discussion of scheduling for tomorrow]
16:29:19 [cygri]
RRSAgent, make logs public
16:30:04 [Zakim]
16:32:11 [pchampin]
8 rue des Marronniers 69002 Lyon - Tel. : 04 72 77 10 00
16:34:05 [manu]
manu has joined #rdf-wg
16:35:04 [Zakim]
disconnecting the lone participant, Rhone_4, in SW_RDFWG(TPACF2F)2:00AM
16:35:07 [Zakim]
SW_RDFWG(TPACF2F)2:00AM has ended
16:35:07 [Zakim]
Attendees were Rhone_4, ScottB, PatH, GavinC, +1.617.838.aaaa, MacTed, gkellogg, AndyS
16:48:38 [AZ]
AZ has joined #rdf-wg
17:03:48 [cygri]
cygri has joined #rdf-wg
17:11:33 [SteveS]
SteveS has joined #rdf-wg
17:17:25 [gkellogg_]
gkellogg_ has joined #rdf-wg
17:50:41 [ivan]
ivan has joined #rdf-wg
17:58:51 [Arnaud]
Arnaud has joined #rdf-wg
18:00:40 [trackbot]
trackbot has joined #rdf-wg
18:01:10 [mox601]
mox601 has joined #rdf-wg
20:26:00 [swh]
swh has joined #rdf-wg
20:36:50 [gkellogg]
gkellogg has joined #rdf-wg
20:41:22 [gkellogg]
gkellogg has joined #rdf-wg
22:06:25 [gkellogg]
gkellogg has joined #rdf-wg
22:28:05 [cygri]
cygri has joined #rdf-wg
22:47:44 [gkellogg]
gkellogg has joined #rdf-wg
23:17:11 [pchampin]
pchampin has joined #rdf-wg
23:21:23 [swh]
swh has joined #rdf-wg
23:33:13 [Arnaud]
Arnaud has joined #rdf-wg
23:46:39 [SteveS]
SteveS has joined #rdf-wg