RDF Data Access Working Group (DAWG) Minutes

22 Feb 2005


See also: IRC log


Yoshio, HowardK, Kevin, AndyS, DaveB, DanC, Kendall_Clark, SteveH, HiroyukiS, PatH, EricP, AlbertoR, JosD



morning all

<Yoshio> evening

<DaveB> DanC: AndyS and I are at HP

Convene, take roll, review records and agenda http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0186.html

<HiroyukiS> P3 is me

<kendall> is the "s" in "sparql" "simple" or "sparql"?

it's sparql (what's diff between 1st and last Kendall?)

<kendall> yoshio: hehe

<DanC> Minutes of DAWG 2005-02-15 telcon for review http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0133.html

minutes accepted as amended

<Yoshio> well, I was absent on 15 (sick in bed)

<DanC> ammendment2: Yoshio Fukushige not at 15 Feb telcon

<Yoshio> thank you

ftf5 Feb 28-1Mar in Cambridge, MA, USA

<DanC> ftf5 agenda

actions 1: all continued except Steve's

<DanC> hm... http://www.w3.org/2001/sw/interest/meetings/tp2005.html $Revision: 1.5 $ of $Date: 2005/02/22 16:02:46 $ by $Author: swick $

ftf5: kendall to scribe

<afs> Boston->CET is 6 hours

dan has requested same phone #


<DanC> steve's offer to host

<DanC> PatH: european SemWeb thing is 1Jun and prev 3 days

<afs> http://www.eswc2005.org/

ftf6: asemantics offer 19-20 may

pat to put offer in writing

<DanC> Yoshio: pls co-locate ftf meeting with other events

yoshio: prefers adjacent to another conference

kevin: no pref

howard: no pref

hiro: no pref

danc: hoping to go to xtech amsterdam

kendall: not much of a pref

steve: prefers later

plus offers to host

pat offered as well

doesn't mind if elsewhere tho

eric: co-locate w/ xtech would be nice

jos: same as alberto

<AlbertoR> alberto: prefers Europe for f2f 6th either Southumpton (or Leiden)

andy: being in Japan 1 week before Leiden would be toughish

dave: also going to japan

<kendall> hmm, there's a good chance I'll be in Japan too, actually.

danc: would like to celebrate candidate rec in June

<DanC> (get there in Apr, celebrate May/Jun)

danc: can we decide in cambridge?

<DanC> so danc's action continues; the rest are done.

SPARQL QL publication

and much happiness all around

eric: operators depending on operand type still a bit unclear

whoops: I missed what was a bit controversial eric?

<DaveB> EBV

<DaveB> from XQuery F&O

thank you

SPARQL result set

<DaveB> DanC mentioned query pubbed around cvs version 1.224/1.225

daveb: action is done

liam's suggest xsl group to review result set spec

daveb: a few comments re attrs vs elements but not significant

<AlbertoR> http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2005Feb/0005.html

<AlbertoR> Damian Steer

daveb: mentions comment and recommendation on xsi:type

<DanC> Re: XML result format and datatypes

<Zakim> DanC, you wanted to note qname "fun"

<ericP> ouch

danc: xsi:type takes qname, with accompanying issues
... comment seems well motivated

<ericP> "1234"^^<http://...myCrazyScheme#thang>

dave: can't turn all uri's into qnames

eric: wsdl uri types are xpointers and end in lparens: can't become qnames

<DanC> (ericp, are those WSDL uris supposed to be URIs for datatypes? or URIs for datatype-related-syntactic-constructs?)

dave will accept action to mull on this

<DanC> ACTION: DaveB to consider use of xsi:dataType ala comment from Steer [recorded in http://www.w3.org/2005/02/22-dawg-minutes#action01]

best practises doc discusses this issue

<ericP> <foo xmlns:foo="http://...myCrazyScheme#" xsi:datatype="foo:thang">1234</foo>

daveb: wants to also think over yes/no in result set

<afs> Eric - that complicates things :-(

<ericP> orig datatype was http://...myCrazyScheme#thang. did we communicate that to XSD? or at least RDF people interpretting theXML data?

<patH> the best practices doc is http://www.w3.org/2001/sw/BestPractices/XSCH/xsch-sw-20050127/

daveb: to review question of errors from protocol affecting result set

malformed query to return defined error in result set ?

<DanC> (hmm... HTTP error codes seem exactly right for the "malformed query" case)

kendall: what about soap, rpc etc?

daveb: runtime errors can also occur (??)

<DaveB> runtime errors such as can't load uri, extension function failed, ...

kendall: some http error codes map nicely, but ...

<DanC> 400 Bad Request

<DanC> looking in http://www.w3.org/2001/sw/DataAccess/proto-wd/ ...

<JosD> http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html#sec10.4.1

kendall: has relevant section in protocol spec re 400 bad request
... but not 100% confident of treatment

danc: looks ok to me

valueTesting, XSCH

daveb: hoping for another draft before ftf, get in yes/no and think about schema thing

pat has done detailed review for authors

thinks nice piece of work!

doesn't think 2nd half impacts our work

erics hasn't read pat's comment yet

pat: 1st half has recommendation on id (?) rather than name

technique of using bracketed ending not practical for compatibility reasons

<ericP> (that technique is XPointer, i believe)

didn't fully understand all ramifications tho

ericp: mostly made sense

pat: it's a bit of a mess

danc: would prefer to see test case how it effects our work

ericp: action continued

pat's: done

jos: sorry, no further progress

to be continued

(prior of section 11 testing values)

jos: 2 test cases: one test case no results; 2nd case an implies returning data

in this case fires once

danc: not sure how well fits existing test structure

but engines doing inference before should be discussed somewhere

danc: saying it's ok, it can happen

<DaveB> Andy refers to ' ?A^^xsd:dateTime'

daveb: notes similarity to casting (dave? andy?)


<DanC> (jos, I think it would be interesting to discuss ?A^^xsd:dateTime in public-cwm-talk)

andy: mentions issue of type promotion from xquery

test only gets applied to final type

<DanC> (my wish for a discussion of this sort of inference in one of our docs doesn't seem to have gone anywhere... hmm... I guess that's ok)

<afs> Opne to suggests for a para to insert somewhere Dan.

danc: may a 1000 test cases bloom

<afs> Pref have that text early, not specific to values

eg do strings automatically get promoted to ints?

danc: test case 2 means change to turtle

turtle + literal subjects

<DaveB> test case 1 does not need literal subjects

test case 2 also has triple variable in it

<DaveB> ?A^^xsd:dateTime math:greaterThan ?B^^xsd:dateTime

can't be done

<DanC> "2005-02-15T00:00:00Z" op:dateTime-greater-than "2000-06-00T00:00:00Z".

<DaveB> Tripe pattern section 2.2

<DaveB> RDF-G union V x RDF-U union V x RDF-G union V

danc: would be in favour of extending turtle to handle literal subjects for test suite
... would require changing turtle, but isn't rdf/xml

<DanC> "2005-02-15T00:00:00Z"^^xsd:dateTime op:dateTime-greater-than "2000-06-00T00:00:00Z"^^xsd:dateTime.

pat: would have pedagogic value to do so

ericp: would hesitate to add a new req: literal subjects

<AlbertoR> same here

<afs> +1 - we check and assume no literal subjects

<DaveB> DaveB also redland

sorry: check for what andy?

<afs> Our Graph layer can do it - Our public API doesn't

danc: doesn't see critical mass of support

<afs> (Check for literal subjects)

jos: you can use bnode to express this in rdf/xml

<ericP> what's the range of owl:sameAs ?

<JosD> rdfs:Resource

steve: literal subjects seem like agood idea, but maybe not in the query language

danc: invites jos to submit test

jos prefers to wait a bit due to bandwith issue

<DanC> ACTION: Jos to propose an optional test based on ACTION op:dateTime triple http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0146.html [recorded in http://www.w3.org/2005/02/22-dawg-minutes#action02]


<DaveB> I hope JosD uses the triple: <DanC> "2005-02-15T00:00:00Z" op:dateTime-greater-than "2000-06-00T00:00:00Z".

dan's email prompted a few thoughts

<kendall> so that's not something we shld consider for dawg work?

<DanC> a service description approach to accessingCollections

proposes to postpone the issue

<kendall> good, that's one less thing i have to read soon. :>

at least for this recommendation, no impetus to standardize seen

<DaveB> I don't think anyone has actually made such a service...

<DanC> PROPOSED: to postpone accessingCollections because (a) our not standardizing it doesn't stop anybody from playing (b) none of the extant designs seems sufficiently mature

pat feels nervous about postponing this

7 different implementations will spring into being

w/ accompanying chaos

danc: not so bad
... feels group has given sufficient thought to this

question: postpone this?

<kendall> KC abstaining, as well

hp abstain

<Yoshio> how about giving just informative example?

<DanC> no, hp did not abstain

<afs> HP is happy to postpone the issue to the next WG

yoshio: sorry

whoops, that's me

yoshio: abstain

howard abstain

hiroyuki: can live w/ postpone

w3c: in favour

<DanC> p a a p p a p a p a

alberto: postpone

<DanC> p a a p p a p a p a p

daveb: postpone

danc: question carries

<kendall> boohiss!


Service Descriptions

no motion to adjourn

<DanC> Re: on log:definitiveService

danc ACTION: 0166.thml

andy said yes, not much other response

andy: you can be 'partially definitive'

kendall: inclined to support something similar

<Yoshio> what do you mean by "partically definitive,"andy?

assumes doesn't need more text from danc

danc: still a bit early for a few (who?)

dan's action's done

<DanC> (I heard SteveH, DaveB say "too early")

kendall wants support in order to do so

danc: (1) we can write a test case (for failure)

<DaveB> sounds too premature to me. get these people together to invent a preidcate themselves and experiment

(2) (howard missed that)

action on sparqlParam: ...

ericp: it offers people providing services much more latitude

<afs> Makes writing servlets hard ??

<DaveB> there are 6 http query params in sparql protocol

<afs> (Servlet API has parameter accessors)

eric to mail on simple concatenation

<kendall> fwiw, I preferred some of those query parameters to be HTTP headers...

has already done: ...

dan's action done

next 3 actions kendall's ...

1 continued

<afs> Kendall, doesn't that mess up HTTP?

2 continued

would like to put 3 on agenda cambridge

<kendall> afs: huh?

<ericP> concatonated query: http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0128 option 4 (look for "append the encoded query directoly")

danc: just giving example of construct w/ optional ok

overloading http error code seems bad idea

(prior is kendall)

<afs> and http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0140.html which suggests layering and making client assembly independent

question of adding an http header ...

<SteveH> http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.46

<DanC> 13.1.2 Warnings

seems possibly relevant

also steve's prior url

<ericP> kendall, http extensions: http://www.w3.org/Protocols/HTTP/ietf-http-ext/

ericp: continued to tech plenary

ericp to expedite somewhat

kendall: lab partner has shown soap protocol, 70% soap binding for sparql protocol

offered to help co-write note on this

what to do?

will try to incorporate under pressure to publish expeditiously

danc: if they joined wg, they could present ideas directly

kendall: odds to join are low, not their priority

kendall action to followup with email

should other bindings be incorporated in protocol spec

ericp: would be nice to leverage off their work

<DanC> (just about out of time)

kendall: thinks available in open source

danc: motion to adjourn

Summary of Action Items

[NEW] ACTION: DaveB to consider use of xsi:dataType ala comment from Steer [recorded in http://www.w3.org/2005/02/22-dawg-minutes#action01]
[NEW] ACTION: Jos to propose an optional test based on ACTION op:dateTime triple http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JanMar/0146.html [recorded in http://www.w3.org/2005/02/22-dawg-minutes#action02]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.111 (CVS log)
$Date: 2005/02/22 16:02:46 $