IRC log of dawg on 2005-03-01

Timestamps are in UTC.

03:11:16 [AndyS]
AndyS has joined #dawg
03:33:03 [DanC_]
DanC_ has joined #dawg
10:01:49 [DaveB]
DaveB has joined #dawg
10:35:24 [JanneS]
JanneS has joined #dawg
10:47:24 [JanneS]
Good Morning US, Europe goes for lunch
11:17:30 [Zakim]
Zakim has left #dawg
12:12:44 [DanC_]
DanC_ has joined #dawg
12:37:07 [JanneS]
good morning
12:45:15 [ericP]
good morning janne
12:52:25 [JanneS]
hi eric
13:30:55 [Zakim]
Zakim has joined #dawg
13:31:45 [JanneS]
called in minute ago - I'm the 1st participant it appears
13:32:54 [JanneS]
zakim, who's here?
13:32:54 [Zakim]
sorry, JanneS, I don't know what conference this is
13:32:55 [Zakim]
On IRC I see JanneS, DaveB, RRSAgent, ericP, DanC
13:34:43 [JanneS]
I will call again once Boston wakes up
13:36:41 [DanC_]
DanC_ has joined #dawg
13:36:47 [DanC_]
RRSAgent, pointer?
13:36:47 [RRSAgent]
13:36:54 [DanC_]
Zakim, this is dawg
13:36:54 [Zakim]
DanC_, I see SW_DAWG(TP)8:30AM in the schedule but not yet started. Perhaps you mean "this will be dawg".
13:37:00 [DanC_]
Zakim, this will be dawg
13:37:00 [Zakim]
ok, DanC_; I see SW_DAWG(TP)8:30AM scheduled to start 7 minutes ago
13:37:16 [DanC_]
Zakim, call OlympiaC
13:37:16 [Zakim]
ok, DanC_; the call is being made
13:37:17 [Zakim]
SW_DAWG(TP)8:30AM has now started
13:37:18 [Zakim]
13:37:36 [kendall]
kendall has joined #dawg
13:37:55 [DanC_]
agenda + nestedOptionals
13:38:13 [HiroyukiS]
HiroyukiS has joined #DAWG
13:38:13 [DanC_]
agenda + issue: fromUnionQuery
13:38:22 [DanC_]
agenda + issue: protocolRootReferent (ACTION KendallC)
13:38:29 [DanC_]
agenda + Service Descriptions
13:38:39 [DanC_]
agenda + protocol walk-thru
13:38:51 [DanC_]
agenda + Plan for last call
13:39:18 [JosD]
JosD has joined #dawg
13:39:34 [DanC_]
Zakim, OlympiaC holds Jos, AndyS, KendallC, Yoshio, HiroyukiS, EricP, Jack, SteveH, DanC
13:39:34 [Zakim]
+Jos, AndyS, KendallC, Yoshio, HiroyukiS, EricP, Jack, SteveH, DanC; got it
13:40:53 [AndyS]
AndyS has joined #dawg
13:41:04 [Yoshio]
Yoshio has joined #dawg
13:41:13 [Yoshio]
Morning all
13:41:25 [DanC_]
Topic: introductions
13:43:21 [DanC_]
Zakim, who's on the phone?
13:43:21 [Zakim]
On the phone I see OlympiaC
13:43:22 [Zakim]
OlympiaC has Jos, AndyS, KendallC, Yoshio, HiroyukiS, EricP, Jack, SteveH, DanC
13:43:30 [JacekK]
JacekK has joined #dawg
13:43:32 [DanC_]
Zakim, who's on the phone?
13:43:32 [Zakim]
On the phone I see OlympiaC
13:43:33 [Zakim]
OlympiaC has Jos, AndyS, KendallC, Yoshio, HiroyukiS, EricP, Jack, SteveH, DanC
13:43:50 [SteveH]
SteveH has joined #dawg
13:43:56 [JacekK]
zakim, OlympiaC also has BijanP, JordiA
13:43:56 [Zakim]
+BijanP, JordiA; got it
13:44:00 [kendall]
Elias, Lee, Jordie -- all from IBM
13:44:08 [JacekK]
zakim, OlympiaC also has Lee
13:44:08 [Zakim]
+Lee; got it
13:45:01 [DanC_]
13:45:54 [DanC_]
Zakim, take up item 1
13:45:54 [Zakim]
agendum 1. "nestedOptionals" taken up [from DanC_]
13:46:23 [Zakim]
13:46:37 [JanneS]
Zakim, +[IPcaller] is JanneS
13:46:37 [Zakim]
sorry, JanneS, I do not recognize a party named '+[IPcaller]'
13:46:41 [DanC_]
OPTIONS: nested optionals as per
13:46:52 [DanC_]
Zakim, IPcaller is JanneS
13:46:52 [Zakim]
+JanneS; got it
13:47:11 [Lee]
Lee has joined #dawg
13:47:23 [DanC_]
OPTIONS: nested optionals as per , prohibit nested optionals by side-condition on grammar
13:47:40 [JacekK]
discussion overflown from yesterday
13:48:02 [JacekK]
DanC: Straw poll, then I'll pick a winner
13:48:12 [DanC_]
as per wD: 2 to 3
13:49:26 [DanC_]
side condition: 3ish
13:50:17 [JacekK]
side condition means: nested optionals are not allowed in the language
13:50:37 [Yoshio]
+1 to side conditions Yoshio
13:50:45 [EliasT]
EliasT has joined #dawg
13:50:59 [JacekK]
zakim, OlympiaC also has BalajiP
13:50:59 [Zakim]
+BalajiP; got it
13:51:01 [DaveB]
I'll weakly vote to do less coding, i.e. no nested optionals
13:51:53 [Yoshio]
Mitä kuuluu, Janne, (Jänee?)
13:52:47 [JanneS]
Hyv, kiitos!
13:53:31 [jordi_]
jordi_ has joined #dawg
13:53:36 [JacekK]
AndyS: with the syntax we were talking about yesterday, it'd be harder to put the condition in the grammar
13:54:17 [JacekK]
editors' preference: as specified, with deference for allowing things to go forward
13:55:33 [JacekK]
DanC: summary of the discussion on optionals: 5.5 Nested optional blocks, there are reasons to disallow them
13:56:05 [JacekK]
SteveH: nested optionals makes it harder to implement, you have to have code to handle them
13:56:22 [AndyS]
13:57:23 [JacekK]
SteveH: it's tricky to implement
13:57:25 [AndyS]
13:58:08 [ericP]
OPTIONAL { ( ?x vcard:N ?vc )
13:58:08 [ericP]
( ?vc vcard:Given ?gname )
13:58:08 [ericP]
OPTIONAL ( ?vc vcard:Family ?fname ) }
13:58:17 [ericP]
13:58:20 [ericP]
OPTIONAL { ( ?x vcard:N ?vc )
13:58:20 [ericP]
( ?vc vcard:Given ?gname ) }
13:58:20 [ericP]
OPTIONAL { ( ?x vcard:N ?vc )
13:58:20 [ericP]
( ?vc vcard:Given ?gname )
13:58:22 [ericP]
( ?vc vcard:Family ?fname ) }
13:58:32 [ericP]
13:58:38 [AndyS]
13:58:39 [ericP]
13:58:39 [ericP]
OUTER JOIN T as T1 ON (T1.s=T0.s AND T1.p=vcard:N)
13:58:39 [ericP]
OUTER JOIN T as T2 ON (T2.s=T1.o AND T1.p=vcard:Given)
13:58:39 [ericP]
OUTER JOIN T as T3 ON (T3.s=T1.o AND T1.p=vcard:Family)
13:58:39 [ericP]
WHERE T0.p=foaf:name
13:58:45 [JacekK]
ericP: we currently don't have dependency on ordering
13:59:08 [DaveB]
I don't believe that - near optionals I think we do.
13:59:31 [DanC_]
ack andy
13:59:40 [JacekK]
DanC: when we get rid of the optionals, can we end up with illegal query?
13:59:58 [JacekK]
AndyS: with optionals we already have ordering issues
14:00:16 [JacekK]
AndyS: there's examples on the comments list
14:00:18 [bijan]
bijan has joined #dawg
14:00:42 [JacekK]
ericP: SQL mapping might be harder if you force ppl to separate them out
14:00:52 [DanC_]
PROPOSED: nested optionals as per
14:01:11 [JacekK]
SteveH: flattened optionals must be supported anyway
14:01:33 [DanC_]
abstaining: Jos, UMD, Steve
14:01:37 [DanC_]
14:01:49 [DanC_]
dave, do you want to vote? I'm not sure you can
14:02:03 [DaveB]
that's up to you
14:02:07 [DaveB]
I abstrain
14:02:09 [bijan]
Can we rephrase EricP's point that there are SQL mappings that are much harder from the normal form that might be significanly better, more optimized, etc.
14:02:17 [DanC_]
very well. dajobe abstaining
14:02:21 [DanC_]
Zakim, who is on the phone?
14:02:21 [Zakim]
On the phone I see OlympiaC, JanneS
14:02:23 [Zakim]
OlympiaC has Jos, AndyS, KendallC, Yoshio, HiroyukiS, EricP, Jack, SteveH, DanC, BijanP, JordiA, Lee, BalajiP
14:02:37 [DanC_]
Dave, the question of whether you can vote is a question of whether you want to be considered present.
14:02:41 [DanC_]
but it doesn't matter much.
14:02:46 [DanC_]
Zakim, close this agendum
14:02:46 [Zakim]
agendum 1 closed
14:02:47 [Zakim]
I see 5 items remaining on the agenda; the next one is
14:02:48 [Zakim]
2. issue: fromUnionQuery [from DanC_]
14:03:23 [JacekK]
DanC: if you say from URI1, URI2 is that the union, the merge or what?
14:03:25 [DaveB]
DanC: I'm not feeling focused enough here to be considered preent.
14:03:29 [DaveB]
14:06:04 [JacekK]
AndyS: the FROM accepts list but we don't say how
14:06:06 [DanC_]
roger, dave
14:06:40 [JacekK]
DanC: in Helsinki we had three groups of options
14:07:29 [JacekK]
DanC: the URIs were only hints in the first version
14:07:48 [JacekK]
DanC: then: drop FROM/WITH and leave it to the system
14:08:15 [JacekK]
DanC: later: specify what the keywords do
14:08:44 [DaveB]
for me, sparql would be significantly work if it can't get docs from the web. like how document() in xslt is very handy. otherwise it's just a static query language and that's boring
14:08:48 [DaveB]
14:08:53 [DaveB]
14:08:55 [JacekK]
AndyS: the test cases need to specify the graph
14:09:03 [kendall]
+1 daveb
14:09:51 [JacekK]
kendall: without pulling stuff from the web the language would not be so interesting
14:10:16 [JacekK]
kendall: I don't care much, grabbing bits is a protocol issue
14:10:54 [JacekK]
kendall: so maybe we could put this in the protocol only, not in the quey lang
14:11:11 [JacekK]
SteveH: there's always some API
14:11:13 [DanC_]
let's look at this again after some protocol discussion
14:11:15 [DanC_]
Zakim, agenda?
14:11:15 [Zakim]
I see 5 items remaining on the agenda:
14:11:16 [Zakim]
2. issue: fromUnionQuery [from DanC_]
14:11:17 [Zakim]
3. issue: protocolRootReferent (ACTION KendallC) [from DanC_]
14:11:19 [Zakim]
4. Service Descriptions [from DanC_]
14:11:20 [Zakim]
5. protocol walk-thru [from DanC_]
14:11:22 [Zakim]
6. Plan for last call [from DanC_]
14:11:24 [kendall]
dont' have a network or don't hae a protocol? don't you sorta always have loopback?
14:11:57 [JacekK]
kendall: propose to close the issue in the issues list
14:12:11 [AndyS]
Alberto proposed: s/WITH/FROM/ ; s/FROM/FROM NAMED/ on the comments list and elsewhere
14:12:18 [JacekK]
DanC: let's consider that later
14:12:21 [DanC_]
Zakim, take up agendum 4
14:12:21 [Zakim]
agendum 4. "Service Descriptions" taken up [from DanC_]
14:12:29 [DaveB]
I prefered NAMED and I think suggested something like that last year
14:12:46 [AndyS]
ack DaveB
14:13:08 [JacekK]
looking at Kendall's proposal
14:13:41 [JacekK]
DanC: how many ppl have read this over? Came out Saturday Ken's time
14:13:42 [DanC_]
14:15:01 [JacekK]
kendall: did some changes to naming, added invocableOperations ...
14:15:25 [JacekK]
kendall: added saddle:operationTarget to tell me which graphs I can interact with
14:16:00 [JacekK]
kendall: added acceptSerializationFormat
14:16:59 [SteveH]
should include the supported extended functions (&ex:myFunc(?x))
14:17:08 [JacekK]
kendall: this is about data format, not query language serialization
14:17:39 [JacekK]
kendall: nervous about acceptSerializationFormat, redundant to content negotiation
14:18:02 [JacekK]
kendall: another issue - how do you identify the various subsets of OWL
14:18:49 [JacekK]
kendall: invocableOps can contain IDs of the ops in our protocol
14:19:25 [JacekK]
kendall: can vary according to graphs, can choose not to support GetGraph on a huge graph
14:20:29 [JacekK]
SteveH: I had in mind something where you wouldn't have to repeat the properties shared between services
14:21:44 [JacekK]
DanC: we can always add extension properties
14:22:02 [JacekK]
DanC: so I'd only keep those where two people can demonstrate interop
14:22:09 [Yoshio]
q+ to ask if it is just xsd:*** that can be used as the object of "saddle:supportsDatatype"
14:22:34 [JacekK]
DanC: we'd find out what interop is as we'd go
14:23:04 [JacekK]
DanC: story about operationPoint
14:24:16 [JacekK]
kendall: if I know a URI is of type OperationPoint, I can GET it and get the service description
14:24:24 [JacekK]
DanC: is that in spec for this type?
14:24:26 [JacekK]
kendall: yes
14:24:51 [JacekK]
kendall: GetSvcDescription would be a required protocol operation
14:25:03 [JacekK]
AndyS: in HTTP, it would be GET
14:25:26 [AndyS]
(that was a question to Kendall for clarification)
14:25:52 [JacekK]
kendall: story about OpTarget
14:26:13 [JacekK]
kendall: on a URI of this type, I'd have to say OPTIONS, not GET because here GET is GetGraph
14:26:26 [bijan]
14:27:21 [JacekK]
DanC: I might need to point from target to service
14:27:58 [DanC_]
ack Yoshio
14:28:46 [JacekK]
Yoshio: is the object of supportedDatatype an XML Schema thing? Can use add their datatypes?
14:28:59 [JacekK]
kendall: it's RDF, put there anything
14:29:30 [JacekK]
DanC: you can put whatever datatype you want there
14:30:03 [JacekK]
Yoshio: can you also describe the ops on the datatypes? Should it be specified as an ontology thing?
14:30:13 [JacekK]
kendall: sounds like a good idea
14:30:27 [JacekK]
kendall: my version doesn't have that (yet)
14:30:44 [JacekK]
bijan: what's the interaction between supported datatypes and the ontology used?
14:31:24 [afs]
afs has joined #dawg
14:31:45 [JacekK]
DanC: I was thinking of a class OperationPoint, all you know is you can perform SPARQL queries, you cannot get a svc description
14:32:46 [JacekK]
bijan: do you want a class that has the QL implicit in it?
14:33:24 [JacekK]
kendall: we could put the class in the SPARQL namespace, making the QL implied
14:33:43 [JacekK]
bijan: trying to understand the semantics of the class here
14:34:23 [JacekK]
bijan: so it would be equivalent to a generic OpPoint with a queryLanguage set to SPARQL
14:35:05 [JacekK]
kendall: I want an OpPoint that identifies a SPARQL service
14:35:18 [JacekK]
kendall: I'd like GetDescription required
14:36:18 [JacekK]
bijan: what does OPTIONS do on OpPoint? Is it the same as GET there?
14:36:37 [bijan]
14:36:42 [Yoshio]
14:36:48 [JacekK]
DanC: we may have to go back to the protocol stuff
14:37:33 [JacekK]
kendall: is the "keeping stuff if we get interoperable pairs" dropped?
14:37:42 [JacekK]
DanC: tried one, got into a lot of issues...
14:38:18 [DanC_]
14:38:31 [JacekK]
DanC: moving to protocol description
14:38:41 [kendall]
for the record, i thought it was working fine. :>
14:38:46 [JacekK]
DanC: my proposal: took example from UCs and Reqs
14:39:23 [JacekK]
DanC: querying for a weblog of whoever
14:39:35 [JacekK]
DanC: then I describe to the encoding
14:40:25 [JacekK]
DanC: touched FROM/WITH slightly
14:41:09 [JacekK]
DanC: warnings can go in Warning header
14:41:24 [JacekK]
kendall: that header is apparently only about caching
14:41:53 [JacekK]
kendall: maybe we could define a specific warning code
14:42:17 [kendall]
i didn't say that. :>
14:42:22 [JacekK]
DanC: and concluding with limiting the results
14:43:22 [JacekK]
bijan: fujitsu labs won't do DAWG without SOAP
14:43:41 [JacekK]
bijan: your document, if it replaces the current protocol, doesn't say anything about SOAP
14:43:57 [ericP]
14:43:58 [JacekK]
bijan: the current protocol doc specifies an abstract protocol
14:44:37 [JacekK]
ericP: a WSDL description could take advantage of the WSDL mechanism for putting things in query
14:45:13 [JacekK]
DanC: the WSDL has been out there for a bit, it describes a GET, not a POST, right?
14:45:30 [JacekK]
DanC: in SOAP, would it be SOAP response or would be SOAP req/resp?
14:46:12 [JacekK]
DanC: can you show me the HTTP that comes out of the fujitsu's binding?
14:46:37 [JacekK]
ericP: it would have POST
14:46:44 [JacekK]
DanC: it's important to have a GET binding
14:46:48 [ericP]
<binding name='httpSparqlBinding'
14:46:48 [ericP]
14:46:48 [ericP]
14:46:48 [ericP]
<operation ref='sl:query'
14:46:48 [ericP]
whttp:method="GET" >
14:46:51 [ericP]
14:46:53 [ericP]
14:46:56 [ericP]
14:47:12 [ericP]
<binding name='httpSparqlBinding'
14:47:12 [ericP]
14:47:12 [ericP]
14:47:12 [ericP]
<operation ref='sl:query'
14:47:12 [ericP]
soap:method="@@@" >
14:47:15 [ericP]
14:47:17 [ericP]
14:47:33 [JacekK]
DanC: we don't have a requirement to give a SOAP interface
14:48:18 [JacekK]
bijan: I'd still like to make it easy to hook in the SOAP binding
14:48:36 [JacekK]
bijan: a normative WSDL would work for me (with HTTP binding) because I can add a SOAP binding easily
14:49:14 [kendall]
q+ to ask about protocol *requirements*
14:49:51 [JacekK]
DanC: the request for WSDL has come from a number of places, begins to look like a requirement
14:50:36 [JacekK]
bijan: story why WSDL is useful: fujitsu does large-scale integration, their middleware uses RDF & OWL, they have a protocol that allows querying
14:50:57 [JacekK]
bijan: they want to interoperate with everything that uses Web services
14:51:08 [JacekK]
bijan: all the MS Office operations now support SOAP
14:51:32 [JacekK]
bijan: so they want to be able to call uniformly SPARQL or Word
14:51:43 [kendall]
14:52:00 [JacekK]
jordi_: doing HTTP methods is less supported than arbitrary WSDL operations
14:52:08 [DanC_]
OPTIONS for stuff to add to UC&R: new requirement: WSDL specification of SPARQL protocol. new use case
14:52:40 [JacekK]
bijan: it's the abstract WSDL that's important to them
14:53:03 [JacekK]
DanC: I'm confident about adding this requirement
14:53:13 [JacekK]
DanC: anybody objects?
14:53:23 [JacekK]
DanC: any volunteers?
14:53:53 [DanC_]
ack kendall
14:53:53 [Zakim]
kendall, you wanted to ask about protocol *requirements*
14:54:06 [JacekK]
bijan and ericP
14:54:28 [JacekK]
kendall: do we have requirement for HTTP?
14:54:32 [JacekK]
DanC: in the charter
14:54:42 [JacekK]
kendall: same strength as SOAP
14:56:10 [timbl]
timbl has joined #dawg
14:56:40 [JacekK]
AndyS: we have requirement for addressable query results
14:56:52 [DanC_]
AFS notes
14:58:15 [AndyS]
Dan notes it's an design objective
14:58:18 [JacekK]
bijan: I'll take the action to write up the req and UC
14:58:32 [DanC_]
PROPOSED: requirement: A WSDL description shall be included in the protocol specification
14:59:44 [JacekK]
DanC: we need a good story for this requirement
15:00:05 [JacekK]
nobody opposed to adding the req, two abstentions
15:00:20 [DanC_]
abstain: AFS, JosD
15:00:26 [DanC_]
15:00:59 [DanC_]
ACTION Bijan: propose text (story? etc.) to support WSDL requirement
15:01:36 [DanC_]
ACTION EricP: review WSDL text proposal
15:02:46 [DanC_]
Bijan: should the story mention or not mention specific companies?
15:02:54 [DanC_]
DanC: small leaning toward specifics...
15:03:03 [DanC_]
Kendall: other stories are "sanitized"
15:03:13 [DanC_]
DanC: if you mention one company, there's some obligation to mention others
15:04:12 [bijan]
Fujistu WSDL:
15:04:18 [JacekK]
ericP: explains the current WSDL description
15:04:25 [bijan]
(It's WSDL 1.1/what ever MS uses at the moment)
15:04:41 [JacekK]
ericP: it's only for query operation
15:04:48 [JacekK]
ericP: adding SOAP binding would be trivial
15:05:15 [JacekK]
ericP: we could get SOAP over email, SOAP over JMS
15:05:46 [DanC_]
15:06:14 [JacekK]
bijan: fujitsu doesn't need to expose the services externally, but internally they have all sorts of different protocols
15:06:14 [ericP]
15:06:49 [JacekK]
ericP: I wrote up a description of the probocol based on the WSDL
15:07:08 [JacekK]
ericP: also explained it to people who don't speak WSDL
15:07:37 [JacekK]
ericP: the URI encoding is like CGI parameters
15:07:53 [jordi_]
q+ to mention that SOAP interoperability is not a given..
15:08:06 [JacekK]
ericP: the goal was to have WSDL but also an easy description for ppl who don't use WSDL
15:08:33 [JacekK]
ericP: it doesn't have an abstract protocol, only on the level of WSDL
15:09:00 [JacekK]
ericP: it has parameters and normative binding to HTTP
15:09:23 [JacekK]
DanC: it seems it has an abstract protocol specified in WSDL interface
15:10:27 [JacekK]
bijan: the document is confusing to me because the textual version is not about the abstract protocol but about the binding
15:11:20 [AndyS]
q+ to aks if there are any non-WSDL cases we need to allow for
15:11:25 [JacekK]
DanC: if we took this document and added it to kendall's doc, would it replace the abstract notation?
15:11:31 [JacekK]
DanC: would the abstract notation go away?
15:12:23 [JacekK]
bijan chairing until break
15:12:28 [JacekK]
chair declares break
15:13:20 [JacekK]
reconvening at 10:30
15:36:34 [Yoshio]
q+ to ask a general/off the discource question about privacy and security: would it be possible to add such information with current framework?
15:36:54 [AndyS]
15:37:43 [timbl]
timbl has joined #dawg
15:38:43 [DanC_]
(break seems to be extending to 10:40)
15:40:49 [JacekK]
chair back to DanC
15:41:44 [JacekK]
DanC: going through kendall's draft of protocol
15:42:08 [JacekK]
DanC: skipping types, we'll get that from WSDL
15:42:34 [JacekK]
DanC: let's go through the operations
15:42:43 [JacekK]
DanC: query
15:45:07 [JacekK]
danC shows the WSDL in his emacs
15:46:48 [Lee]
Lee has left #dawg
15:46:50 [Yoshio]
q+ to ask if it is possible to add optional (OP-dependant) argument,?
15:47:01 [Lee]
Lee has joined #dawg
15:47:41 [JacekK]
ericP: I didn't meen to supercede kendall's WSDL
15:47:49 [JacekK]
kendall: mine's more minimal
15:50:35 [JacekK]
danc editing the WSDL on screen
15:54:00 [JacekK]
bijan: operation point is WSDL endpoint
15:55:39 [JacekK]
discussion where operation target is - query language or protocol
15:55:50 [JacekK]
AndyS: I like it in the query language for scripting purposes
15:56:08 [JacekK]
AndyS: have no problem with duplication
15:56:32 [JacekK]
kendall: problem with clashes if it's in both places
15:56:51 [patH]
patH has joined #dawg
15:57:15 [JacekK]
DanC: let's start with how to do this in the protocol
15:59:12 [JacekK]
bijan: we could have the target in an attribute
15:59:25 [JacekK]
JacekK: probably not because the operation follows the WSDL URI operation style
16:06:33 [JacekK]
DanC: why do we have getGraph?
16:06:52 [JacekK]
kendall: for retreival operation over a different protocol than HTTP
16:11:38 [JacekK]
kendall: getGraph is an equivalent of the HTTP GET for when you aren't on HTTP
16:11:53 [JacekK]
AndyS: this seems very useful but not tied to SPARQL
16:12:49 [JacekK]
AndyS: this is "get the whole thing" which is in no sense specific to SPARQL
16:13:22 [JacekK]
JacekK: we have WS-Transfer for a GET operation
16:13:51 [JacekK]
AndyS: so I don't want this in this particular namespace of working group
16:13:59 [JacekK]
16:14:39 [JacekK]
DanC: getGraph doesn't have critical mass of support
16:15:02 [JacekK]
kendall: we need an example in the ql for getting the full graph
16:15:06 [SteveH]
CONSTRUCT * WHERE GRAPH <uri> (?x ?y ?z)
16:15:44 [AndyS]
Not quite
16:16:04 [DanC_]
ACTION AndyS: explain how to get a whole graph with CONSTRUCT * and GRAPH.
16:16:14 [AndyS]
No need for GRAPH (always) because it may be where it was sent.
16:16:22 [DanC_]
ACTION KendallC: refer to "get the whole graph" explanation from protocol spec
16:16:40 [AndyS]
Alos - no clear (:-) that is what CONSTRUCT * ... GRAPH means in general
16:16:43 [JacekK]
DanC: getServiceDescription
16:17:37 [JacekK]
ericP: looks like the namespace document problem
16:19:10 [JacekK]
JacekK: discovery is already handled in the protocols (HTTP OPTIONS, SOAP WS-MetadataExchange)
16:19:23 [JacekK]
bijan: my users like to have this in the protocol
16:19:52 [JacekK]
AndyS: two views: 1) it's not a requirement, nothing to do with us, 2) we'd like to have that
16:20:11 [JacekK]
AndyS: if we do in fact do it, people who we don't satisfy might go away
16:20:58 [Yoshio]
Isn't it of some value if we get the result (WSD) in a RDF graph?
16:21:01 [JacekK]
bijan: on behalf of my users, can I ever push a potentially useful thing that "might drive ppl away"?
16:22:51 [JacekK]
DanC: adding interface SPARQLDiscovery
16:24:19 [JacekK]
DanC: we could at this point make a WG decision and action ppl to figure out the details
16:26:07 [JacekK]
DanC: I'd rather not talk about faults now, it can be figured out by the ppl figuring out the details
16:26:36 [DanC_]
PROPOSED: that the SPARQL WSDL description shall have 2 interfaces (SPARQLQuery and SPARQLDiscovery), each with one operation
16:27:35 [ericP]
16:27:47 [DanC_]
chair detects insufficient support just yet
16:27:51 [JacekK]
SteveH: I may support this if somebody explains it to me and I implement it
16:28:34 [DanC_]
ACTION KendallC: add WSDL description of protocol to editor's draft, propose to WG
16:28:37 [tlr]
tlr has joined #dawg
16:29:00 [JacekK]
DanC: can we conclude this discussion at this moment?
16:29:25 [RRSAgent]
16:29:33 [patH]
remind me of the zakim dial-in code for this session? Ta.
16:29:37 [JacekK]
AndyS: is getServiceDescription mandatory for conformance with SPARQL?
16:29:42 [tlr]
zakim, code?
16:29:42 [Zakim]
the conference code is 3294 (tel:+1.617.761.6200), tlr
16:30:18 [DanC_]
16:30:51 [patH]
apparetnly the conference is restricted at this time. Did you guys go over the time limit??
16:31:00 [JacekK]
jordi_: I expect that we would have a normative SOAP binding, too
16:31:31 [JacekK]
jordi_: but optional
16:32:08 [DanC_]
Zakim, passcode?
16:32:08 [Zakim]
the conference code is 3294 (tel:+1.617.761.6200), DanC_
16:32:39 [DanC_]
Zakim, room for 3?
16:32:40 [Zakim]
ok, DanC_; conference Team_(dawg)16:32Z scheduled with code 83261 (TEAM1) for 60 minutes until 1732Z
16:32:53 [bijan]
patH, we're tryign to resolve
16:32:57 [DanC_]
Zakim, who's on the phone?
16:32:57 [Zakim]
On the phone I see OlympiaC, JanneS
16:32:59 [Zakim]
OlympiaC has Jos, AndyS, KendallC, Yoshio, HiroyukiS, EricP, Jack, SteveH, DanC, BijanP, JordiA, Lee, BalajiP
16:33:13 [patH]
ok, thanks. dont waste too much time on it.
16:33:22 [DanC_]
janne, we're switching to 83261, ok?
16:33:30 [JanneS]
16:33:37 [DanC_]
Zakim, call OlympiaC
16:33:37 [Zakim]
16:33:38 [Zakim]
ok, DanC_; the call is being made
16:33:39 [Zakim]
16:34:09 [DanC_]
(and Zakim is wicked cool.)
16:34:22 [DanC_]
Zakim, what conference is this?
16:34:22 [Zakim]
this is SW_DAWG(TP)8:30AM conference code 3294
16:34:36 [DanC_]
Zakim, drop OlympiaC
16:34:36 [Zakim]
OlympiaC is being disconnected
16:34:38 [Zakim]
16:34:49 [DanC_]
Zakim, this will be Team_(dawg)
16:34:49 [Zakim]
ok, DanC_; I see Team_(dawg)16:32Z scheduled to start 2 minutes ago
16:34:55 [DanC_]
Zakim, call OlympiaC
16:34:55 [Zakim]
ok, DanC_; the call is being made
16:34:56 [Zakim]
Team_(dawg)16:32Z has now started
16:34:58 [Zakim]
16:35:21 [DanC_]
Zakim, drop OlympiaC
16:35:21 [Zakim]
OlympiaC is being disconnected
16:35:22 [Zakim]
Team_(dawg)16:32Z has ended
16:35:24 [Zakim]
Attendees were OlympiaC
16:35:29 [DanC_]
Zakim, call OlympiaC
16:35:29 [Zakim]
ok, DanC_; the call is being made
16:35:30 [Zakim]
Team_(dawg)16:32Z has now started
16:35:30 [Zakim]
16:35:32 [Zakim]
16:35:33 [Zakim]
16:35:40 [DanC_]
Zakim, who's on the phone?
16:35:40 [Zakim]
On the phone I see OlympiaC
16:35:48 [DanC_]
pat, janne, I think it's safe to dial in now
16:36:04 [DanC_]
Zakim, take up item 4
16:36:04 [Zakim]
agendum 4. "Service Descriptions" taken up [from DanC_]
16:36:26 [JanneS]
16:36:48 [JacekK]
DanC: we have two classes (and one for them combined)
16:37:51 [JacekK]
DanC edits service descriptions on-screen
16:38:13 [JacekK]
DanC: what's the status of WSDL2?
16:38:19 [JacekK]
bijan: going to have a second last call
16:38:27 [JacekK]
AndyS: I'm a bit behind the curve on that
16:40:17 [JacekK]
ericP: but we do not expect any changes relevant to us
16:40:48 [patH]
I still can't dial in, and Zakim seems to have got the stutters. Never mind, I'll follow on IRC.
16:40:56 [kendall]
hi pat ;>
16:41:02 [patH]
Hi Kendall
16:41:15 [JanneS]
me neither, let's do text
16:41:29 [JacekK]
bijan: it'd be useful to have also WSDL 1.1 (perhaps non-normative)
16:46:04 [JacekK]
DanC: any feelings on depending on WSDL RDF mapping?
16:46:14 [JacekK]
bijan: we'll finish that mapping together with WSDL 2
16:46:35 [JacekK]
AndyS: we're getting timing issues on finishing SPARQL
16:46:54 [JacekK]
bijan: but we want dependency on WSDL 2, so RDF mapping doesn't add time slip
16:48:21 [patH]
and of coure 2 == 2.0
16:49:50 [AndyS]
In XQuery WSDL 2 = (WSDL_2.0 WSDL_2.1 WSDL_2.2) by general comparison operation
16:50:05 [JacekK]
DanC: do we need a superinterface to the two we already have? will we have to do transitive closure?
16:50:31 [JacekK]
bijan: the RDF mapping will do that transitive closure itself
16:51:48 [JacekK]
DanC: we need to redecide on the interfaces
16:52:00 [DanC_]
PROPOSED: that the SPARQL WSDL description shall have 3 interfaces (SPARQLQuery and SPARQLDiscovery and SPARQLQueryAndDiscovery), each with one operation
16:52:10 [DanC_]
RESOLVED. same abstentions.
16:53:23 [DaveB]
same abstentions as what?
16:53:35 [JacekK]
as the previous resolved proposal
16:53:45 [JacekK]
where we had only 2 interfaces
16:54:11 [bijan]
16:54:34 [DaveB]
I don't see that recorded....
16:55:01 [JacekK]
DanC: supportsDatatype can also have user datatypes
16:55:15 [bijan]
16:56:18 [JacekK]
abstentions on the proposal: AndyS, JosD
16:57:18 [AndyS]
q+ to ask about assumption of OWL abilities
16:57:34 [JacekK]
discussion about "supportedDatatype"
16:57:46 [JacekK]
bijan: maybe supportedOperations on datatypes would be better
16:58:07 [patH]
actually we might need both of those. They arent the same, quite.
16:58:47 [Yoshio]
and what with ontologies?
16:59:06 [JacekK]
AndyS: RDF doesn't have a way of describing what datatypes are supported, that's an OWL feature
16:59:16 [DanC_]
16:59:18 [JacekK]
bijan: in RDF, you have extensible datatypes, with no way of declaring them
16:59:24 [DanC_]
(lunch in T-1min)
16:59:51 [ericP]
path, do you care to weigh in before we ring off?
17:00:07 [ericP]
(not that i'm chairing)
17:00:36 [JacekK]
AndyS: service description may change, what should it tell me about support of datatypes?
17:00:53 [DanC_]
17:00:57 [JacekK]
DanC: meeting adjourning for lunch
17:01:44 [patH]
well, I just meant that not supported datatype means never heard of the typoe, while not supoprted operation means heard of the type, but not that opeartaion on it.
17:02:00 [patH]
Enjoy your feeding, guys.
17:02:40 [DanC_]
(lunch. resume at 1pm Boston time)
17:02:48 [DanC_]
Zakim, drop OlympiaC
17:02:48 [Zakim]
OlympiaC is being disconnected
17:02:51 [Zakim]
Team_(dawg)16:32Z has ended
17:02:53 [Zakim]
Attendees were Dialer, OlympiaC
17:06:28 [JanneS]
bon appetit
18:00:21 [DanC_]
so... resuming...
18:01:01 [jordi_]
jordi_ has joined #dawg
18:01:11 [Yoshio]
Yoshio has joined #dawg
18:02:32 [EliasT]
EliasT has joined #dawg
18:03:31 [kendall]
kendall has joined #dawg
18:04:27 [DanC_]
Zakim, agenda?
18:04:27 [Zakim]
I see 5 items remaining on the agenda:
18:04:28 [Zakim]
2. issue: fromUnionQuery [from DanC_]
18:04:30 [Zakim]
3. issue: protocolRootReferent (ACTION KendallC) [from DanC_]
18:04:31 [Zakim]
4. Service Descriptions [from DanC_]
18:04:33 [Zakim]
5. protocol walk-thru [from DanC_]
18:04:35 [Zakim]
6. Plan for last call [from DanC_]
18:04:47 [DanC_]
agenda + update/insert/manage
18:05:08 [AndyS]
AndyS has joined #dawg
18:06:23 [DanC_]
Zakim, take up item 2
18:06:23 [Zakim]
agendum 2. "issue: fromUnionQuery" taken up [from DanC_]
18:06:24 [Lee]
Lee has joined #dawg
18:06:40 [HiroyukiS]
HiroyukiS has joined #DAWG
18:07:43 [JosD]
JosD has joined #dawg
18:09:08 [DanC_]
OPTIONS: (a) take FROM/WITH out of QL (b) figure out the interaction between FROM in QL and FROM in proto
18:10:46 [kendall]
taking all from/with out of QL makes the queries slightly less "portable"...
18:11:21 [JosD]
Kendall: what's in the protocol is what server does
18:12:07 [JosD]
.. it further looks in query
18:12:27 [DanC_]
in case of mySvc?query=$query;from=http://ex/data.rdf where $query = SELECT ?x FROM http://exb/datab.rdf ...
18:13:05 [JosD]
Steve: don't buy argument there's no protocol
18:13:37 [Yoshio]
18:16:39 [JosD]
DanC: for how many doe it appeal to take FROM/WITH out of QL? 3-ish
18:16:57 [JosD]
18:17:26 [JosD]
18:17:52 [patH]
18:18:42 [timbl]
timbl has joined #dawg
18:19:13 [JosD]*%2BWHERE%2B%2528%2524book%2Bdc%253Atitle%2B%2524title%2529%250D%250A&language=sparql&
18:19:15 [JosD]
itle%0D%0AWHERE%0D%0A++%28%3Fx+dc%3Atitle+%3Ftitle%29%0D%0A&Go=Go&.cgifields=language does have separate RDF content URIs box
18:19:23 [timbl]
timbl has joined #dawg
18:20:12 [JosD]
for one can put it in query box
18:20:52 [JosD]
Kendall: found Ansy's calendar example quite appealing
18:21:17 [JosD]
chair prefers to have it out of QL
18:22:13 [JosD]
Andy: it basically means taking out
18:22:40 [JosD]
.. actually whole section 9
18:23:17 [JosD]
.. doesn't affect GRAPH keyword
18:23:31 [JosD]
.. was SOURCE before
18:23:55 [DanC_]
WHEREAS we've agreed to have DataSet in our WSDL, PROPOSED: to remove section 9 Combining WITH and FROM from QL.
18:24:54 [Yoshio]
what if one want his query excused against a merge of default graph and particular graphs?
18:25:18 [patH]
Can I vote against the proposal by irc?
18:25:38 [DanC_]
oh, hi pat.
18:25:57 [patH]
Hi dan.
18:26:18 [DanC_]
I'm happy for somebody in the room (say, me) to relay arguments from you
18:26:39 [AndyS]
18:27:04 [patH]
OK. I guess I don't see what is wroing with section 9 (minoir things, but...) but it seeems useful, so whay are we trashing it?
18:27:30 [DanC_]
to move it to the protocol.
18:27:30 [Lee]
The proposal is to transfer it from the language to the protocol, and remove it from the language to avoid dealing with interactions between having it in both places.
18:27:34 [kendall]
q+ to ask about the dependencies
18:27:55 [patH]
Ah, I see. Well, OK as long as language users acan have some control.
18:28:31 [kendall]
path: basically pushing it out of the QL and into some tools API/UI/whatever
18:28:42 [patH]
LIke, how do I write a query directed at a particualr graph?
18:29:12 [Lee]
The GRAPH keyword remains in the QL for that purpose, I believe.
18:29:17 [patH]
Remember, j=kendall, Im a user. I don't expect to even look at APIs
18:29:39 [kendall]
yeah. that's why i said (meant to say) api or user interface...
18:29:55 [JosD]
DanC: Cwm has name for BG graph
18:30:03 [patH]
Hmmm, but then isnt the QL exactly the interface i would expect to be using?
18:30:15 [kendall]
in some cases, i suppose so
18:30:46 [kendall]
[we swap in some test cases for this area...]
18:30:46 [patH]
HOw about we keep it in the QL and say that the protocol should follow the directions in the QL but can take over if trhey are absent?
18:31:01 [patH]
OK, ill look at the cases.
18:31:04 [kendall]
er, s/test/spec/ -- 8.4 example
18:31:20 [patH]
I know this is too slow, take it that my worries are on the table and I wills shut up.
18:31:25 [kendall]
path: yes, that's what i proposed, but there's worries that that makes writing clients more difficult because they have to parse sparql.
18:31:37 [JosD]
18:32:00 [patH]
Im more concerend with users than clients :-) We have too many software writers on these WGs.
18:32:08 [kendall]
patH: also, if from/with are in ql and protocol, we have to specify some way to resolve conflicts
18:32:21 [JosD]
i.e. 8.4 GRAPH and a background graph
18:32:32 [patH]
I did that above. QL has priority.
18:33:00 [kendall]
well, some of us thing protocol does :>
18:33:19 [kendall]
but perhaps there's an axiom about QL expressing user intent more reliably or something?
18:33:29 [Lee]
patH: Does the GRAPH keyword not do what you're looking for as a user of the QL?
18:33:34 [patH]
Axiom, right. ;>
18:33:46 [ericP]
related thread
18:33:56 [kendall]
Lee: he wants it *all* in the QL, not just addressing a particular named graph via GRAPH -- er, I think. :>
18:33:58 [patH]
Lee: maybe.
18:34:34 [patH]
Well, if Im in a minority of 1 I will just abstain and go with the flow.
18:34:45 [kendall]
but yr not. :>
18:34:52 [kendall]
it's seems about 50-50ish
18:34:57 [JosD]
test case: SELECT ?a WHERE GRAPH ?g {:a :b :c} {?g dc:source ?a}
18:35:28 [kendall]
we've switched to working on the semantics, which is harder about which channels it's communicated on
18:35:35 [DaveB]
DaveB has joined #dawg
18:35:35 [kendall]
18:35:41 [kendall]
hi dave
18:35:49 [JosD]
input: bg: :g1 dc:source :src
18:35:53 [patH]
18:36:07 [JosD]
named g1: :a :b :c
18:36:53 [JosD]
this test case versus query ASK {:a :b :c}
18:42:24 [timbl]
18:42:34 [timbl]
q+ andy
18:43:22 [kendall]
18:43:25 [kendall]
18:43:54 [JosD]
better versus SELECT * WHERE {:a :b :c}
18:44:03 [AndyS]
q- andy
18:44:04 [JosD]
.. should give emty answer
18:46:26 [DanC_]
(I don't understand 2/3rds of this. I tried to ground it in a test case. I failed. I could use chairing help)
18:47:30 [kendall]
but, dan, that means you grok 1/3rd, which is 50% more than me! :>
18:47:51 [ericP]
seems to be self-chairing
18:49:52 [JosD]
18:49:59 [patH]
I have to admit it is hard to follow using IRC :>
18:50:15 [kendall]
yes, it's a bit ramblely, so hard to scribe for irc.
18:53:45 [ericP]
prediated trust example in
18:55:14 [JosD]
RDF dataset is at issue here
18:55:31 [Yoshio]
the background one?
18:58:06 [patH]
Seems to me that trust discussions are muddying the water. Part of the idea of named graphs was to name a graph, not a graph resource (= time-series of graphs). Maybe we need to be more clear on this distinction.
18:58:29 [AndyS]
The whole design - does the background graph always includes all triples from the named graphs?
18:58:43 [patH]
We use URIs for both, I guess is the cnetral issue. Hmm.
18:59:21 [patH]
Andy, I thought not.
18:59:47 [AndyS]
I was answring Yoshio
18:59:55 [Yoshio]
q+ to ask what will be bound for GRAPH ?src {<a> <b> <c>} if that triple is found only in the background graph which is not (yet?) named.
18:59:57 [AndyS]
I agree trust is muddying
19:00:28 [JosD]
19:00:46 [patH]
what is answer to yoshio?
19:00:59 [DanC_]
yoshio, which design is your question about? the one in the editor's draft?
19:01:01 [AndyS]
"The design ..."
19:01:43 [Yoshio]
dan: the one in the editor's draft and the current one in discussion, which I can't follow
19:02:07 [DanC_]
ack yoshio
19:02:07 [Zakim]
Yoshio, you wanted to ask what will be bound for GRAPH ?src {<a> <b> <c>} if that triple is found only in the background graph which is not (yet?) named.
19:02:36 [DanC_]
WHEREAS we've agreed to have DataSet in our WSDL, PROPOSED: to remove section 9 Combining WITH and FROM from QL.
19:02:52 [timbl]
q+ to suggest a closely related test case. SELECT ?x WHERE GRAPH ?g <a> <b> <c>
19:03:54 [DanC_]
ack timbl
19:03:54 [Zakim]
timbl, you wanted to suggest a closely related test case. SELECT ?x WHERE GRAPH ?g <a> <b> <c>
19:04:08 [timbl]
q+ to suggest a closely related test case. SELECT ?g WHERE GRAPH ?g <a> <b> <c>
19:04:16 [DanC_]
ack timbl
19:04:16 [Zakim]
timbl, you wanted to suggest a closely related test case. SELECT ?g WHERE GRAPH ?g <a> <b> <c>
19:04:32 [timbl]
19:04:36 [timbl]
19:04:54 [timbl]
graph1234567890234567895678: <a> <b> <c>.
19:05:18 [patH]
Dan, nem con from me on the proposal as I now understand it.
19:05:37 [kendall]
path: ?
19:05:49 [patH]
sorry, no objection. abstain.
19:07:33 [patH]
patH is now an ex-participant. He has gone to meet his lunch.
19:07:57 [JosD]
q+ to say tath we implemented via --graph g1234 asserts all it's triples as <g1224> q:graph {:a :b :c}.
19:08:58 [DanC_]
ack josd
19:08:58 [Zakim]
JosD, you wanted to say tath we implemented via --graph g1234 asserts all it's triples as <g1224> q:graph {:a :b :c}.
19:10:22 [DanC_]
WHEREAS we've agreed to have DataSet in our WSDL, PROPOSED: to remove section 9 Combining WITH and FROM from QL.
19:10:51 [DanC_]
WHEREAS we've agreed to have DataSet in our WSDL, PROPOSED: to remove section 9 Combining WITH and FROM from QL; i.e. remove WITH/FROM syntax
19:11:27 [JosD]
appeals to 3 people
19:11:54 [DanC_]
abstaining: AFS, Yoshio, HiroyukiS
19:12:00 [DaveB]
I'm against (but I'm not voting)
19:12:04 [DanC_]
19:12:16 [DanC_]
19:12:16 [JosD]
RESOLVED: to remove section 9 Combining WITH and FROM from QL; i.e. remove WITH/FROM syntax
19:12:39 [DanC_]
close agendum 5
19:13:36 [Yoshio]
q+ to ask again the same question now (may I?)
19:13:48 [DanC_]
Zakim, take up agendum 3
19:13:48 [Zakim]
agendum 3. "issue: protocolRootReferent (ACTION KendallC)" taken up [from DanC_]
19:14:02 [DanC_]
PROPOSED: that doing WSDL addresses protocolRootReferent
19:16:35 [DanC_]
abstaining: KC, SH
19:16:37 [DanC_]
19:16:43 [DanC_]
Zakim, close this agendum
19:16:43 [Zakim]
agendum 3 closed
19:16:44 [Zakim]
I see 4 items remaining on the agenda; the next one is
19:16:45 [Zakim]
2. issue: fromUnionQuery [from DanC_]
19:16:55 [DanC_]
Zakim, fix issue fromUnionquery
19:16:56 [Zakim]
I don't understand 'fix issue fromUnionquery', DanC_
19:16:57 [DanC_]
19:17:43 [DanC_]
Zakim take up agendum 4
19:17:50 [DanC_]
Zakim, take up agendum 4
19:17:50 [Zakim]
agendum 4. "Service Descriptions" taken up [from DanC_]
19:18:10 [DanC_]
SH: service descriptions around datatypes seems immature, on second thought
19:20:29 [JosD]
SH: wannted to add support functions ala _:aGeoSvc sl:extensionFunctions geo:distance.
19:26:33 [AndyS]
The URI for RDQL is
19:27:45 [JosD]
Tim: a languague subset of b language versus their specification documents
19:28:16 [SteveH]
SteveH has joined #dawg
19:37:55 [JosD]
_:evolvingSvc saddle:queryLanguageSpec <>.
19:45:31 [JosD]
versus _:evolvingSvc saddle:queryLanguage [ saddle:spec <>; ...].
19:50:43 [Yoshio]
q+ to asjk if the service uses some extra inference
19:51:18 [JosD]
_:factbookService saddle:dataSet <a>, <b>.
19:56:44 [JosD]
_socialNetworkingService saddle:vocabulary <v>.
20:06:21 [JosD]
=== break time
20:06:57 [DanC_]
(break until 3:20 local time)
20:19:30 [patH]
patH has joined #dawg
20:22:13 [DanC_]
(resuming from break, a little slowly...)
20:22:17 [DanC_]
Zakim, agenda?
20:22:17 [Zakim]
I see 4 items remaining on the agenda:
20:22:18 [Zakim]
2. issue: fromUnionQuery [from DanC_]
20:22:20 [Zakim]
4. Service Descriptions [from DanC_]
20:22:22 [Zakim]
6. Plan for last call [from DanC_]
20:22:24 [Zakim]
7. update/insert/manage [from DanC_]
20:22:54 [EliasT]
EliasT has joined #dawg
20:23:09 [Yoshio]
20:27:19 [DanC_]
agenda + WSDL mapping input
20:28:22 [EliasT]
EliasT has joined #dawg
20:31:51 [DanC_]
ACTION Kendall: incorporate service description discussion notes in protocol spec
20:34:16 [timbl]
q+ to suggest that closure be a buinary relation between graphs
20:35:07 [DanC_]
ack yoshio
20:35:07 [Zakim]
Yoshio, you wanted to ask again the same question now (may I?) and to asjk if the service uses some extra inference
20:36:27 [timbl]
q+ to suggest that closure be a buinary relation between graphs: _ukCSLit saddle:database ex:closure. foo.rdf rdfs:closure ex:closure.
20:37:22 [ericP]
q+ to suggest sepparation between soundness and completeness (in deference to practicality)
20:39:18 [patH]
zakim, code?
20:39:18 [Zakim]
the conference code is hidden, patH
20:39:45 [ericP]
pat, getting chairs attention...
20:40:10 [bijan]
bijan has joined #dawg
20:40:35 [DanC_]
er... I think our telcon res is over. I guess I could summon another
20:40:47 [Jacek]
Jacek has joined #dawg
20:40:59 [patH]
Ok, no matter.
20:42:24 [patH]
I second ericP's suggestion to separaate soundness and completeness.
20:42:51 [Yoshio]
pat: my question was : why not coin other name for closedOver to generalize it to cover for the cases of inferred graphs
20:43:09 [patH]
Also agree with timbl that closure can be expressed as binary relaionbetwen graphs. In fact its owl;functional, right?
20:43:13 [ericP]
i think we don't want to know whether it is an inferred graph
20:43:27 [kendall]
bijan explaining how this is different (?) in owl.
20:43:29 [ericP]
i think we should dial you in
20:43:36 [ericP]
path, are you invested in this?
20:43:53 [patH]
Im interested, for sure. Maybe I can dial direct?
20:44:00 [ericP]
zakim, space for 3?
20:44:01 [Zakim]
ok, ericP; conference Team_(dawg)20:44Z scheduled with code 83261 (TEAM1) for 60 minutes until 2144Z
20:44:08 [Yoshio]
ericP: so I want to generalize the idea of background graph to inferred one
20:44:08 [kendall]
pat, can you do audio ichat?
20:44:22 [patH]
I don't know how...
20:44:29 [kendall]
have a mac?
20:44:39 [ericP]
zakim, please dial OlympiaC
20:44:39 [Zakim]
ok, ericP; the call is being made
20:44:40 [Zakim]
Team_(dawg)20:44Z has now started
20:44:41 [Zakim]
20:44:43 [kendall]
eh, eric's got it
20:45:00 [ericP]
please dial zakim, pat
20:45:07 [ericP]
20:45:11 [patH]
20:45:20 [Zakim]
20:45:21 [timbl]
Zakim, what is the passcode?
20:45:21 [Zakim]
the conference code is 83261 (tel:+1.617.761.6200), timbl
20:46:27 [kendall]
we're moving white board pat
20:46:51 [ericP]
[hijinks ensue]
20:48:07 [DanC_]
steveH has a "computer science lit" service... it's "close over RDFS" in some sense. we're considering:
20:48:20 [AndyS]
SELECT ?x WHERE { ?x :loves ?y }
20:48:25 [ericP]
SELECT ?x WHERE ?x loves ?y
20:48:32 [DanC_]
_:ukLitSvc saddle:closedOver rdfs:semantics.
20:48:55 [ericP]
WHERE ?x a Philanderer
20:49:39 [DanC_]
e.g. input: :bob loves: [ a [ owl:unionOf (:Students :Faculty) ] ].
20:54:03 [DanC_]
_:bioSvc :deductiveClosure @@:owl. #???
20:54:14 [ericP]
20:54:19 [ericP]
20:54:45 [DanC_]
_:bioSvc :dataSet [ :deductivelyClosedUnder @@:owl].
20:57:37 [ericP]
20:58:12 [ericP]
timbl, i you spoke after you queued yourself. did you cover your queued message?
20:58:48 [DanC_]
SELECT ?svc WHERE ... # how to find steve's service
20:59:21 [kendall]
case 1: only match against "told" triples
20:59:26 [DanC_]
BP: yes, we're building clients that discriminate on server inference capability, ala steve's cs lit
20:59:30 [kendall]
case 2: match against something that uses some OWL semantics
20:59:46 [kendall]
case 3: match against the deductive closure
20:59:54 [kendall]
path: is that roughly right?
21:01:27 [DanC_]
q+ to try to kinda wind up
21:01:38 [timbl]
21:01:38 [DanC_]
ack timbl
21:02:28 [AndyS]
DanC summarise process - wants concrete proposal by action
21:03:42 [DanC_]
ACTION Bijan: work on "closeOver" work-alike with SteveH
21:03:55 [DanC_]
Zakim, agenda?
21:03:55 [Zakim]
I see 5 items remaining on the agenda:
21:03:56 [Zakim]
2. issue: fromUnionQuery [from DanC_]
21:03:58 [Zakim]
4. Service Descriptions [from DanC_]
21:04:00 [Zakim]
6. Plan for last call [from DanC_]
21:04:02 [Zakim]
7. update/insert/manage [from DanC_]
21:04:04 [Zakim]
8. WSDL mapping input [from DanC_]
21:04:41 [DanC_]
Zakim, close agendum 4
21:04:41 [Zakim]
agendum 4 closed
21:04:42 [Zakim]
I see 4 items remaining on the agenda; the next one is
21:04:44 [Zakim]
2. issue: fromUnionQuery [from DanC_]
21:04:54 [AndyS]
. ACTION PaulC: get an action item
21:05:08 [AndyS]
XSLT/XQUERY to LC on April 4
21:05:25 [DanC_]
PaulC: XQuery WG agreed to go to last call 4 April
21:05:41 [DanC_]
Zakim, take up agendum 6
21:05:41 [Zakim]
agendum 6. "Plan for last call" taken up [from DanC_]
21:07:54 [AndyS]
DanC reviews WG schedule
21:08:34 [AndyS]
LC was to be 17 March but things have changed
21:10:21 [AndyS]
Possibility 1: End March
21:10:30 [patH]
Did I miss anything important?
21:10:39 [ericP]
don't think so
21:10:39 [AndyS]
DanC: LC min duration is 3weeks
21:11:27 [AndyS]
DanC: there shoudl be input from other WGs ( Dan/chair askes them for input )
21:12:07 [AndyS]
DanC: pref all docs to LC etc together
21:13:31 [AndyS]
DanC: but not necessary
21:13:52 [AndyS]
Discussion: should protocol be after query language?
21:14:05 [AndyS]
Jos: stage testcases as well
21:18:01 [DanC_]
Chair gets some advice in favor of putting sparql punctuation syntax on the issue lists
21:18:14 [DanC_]
Chair is considering acknowledging ServiceDescription as an issue too
21:19:02 [Yoshio]
q+ to ask if there's room in our result format and protocol for user/service to embed privacy/security information (in future) ... If so, why not show an example to ppl out in the world, if not, should we make the room? (postpone?)
21:19:23 [DanC_]
BP: maybe XML Schema connection to VBR on issues list?
21:20:29 [DanC_]
ack yoshio
21:20:29 [Zakim]
Yoshio, you wanted to ask if there's room in our result format and protocol for user/service to embed privacy/security information (in future) ... If so, why not show an example
21:20:32 [Zakim]
... to ppl out in the world, if not, should we make the room? (postpone?)
21:22:29 [AndyS]
e.g. results returned encrypted
21:22:54 [AndyS]
(not clear about partial parts of the results)
21:23:09 [DanC_]
ACTION KendallC: add "since we're using WSDL, you can use WS-Policy..." to protocol draft
21:24:59 [DanC_]
DanC: consider adding 'privacy/security' issue, even if we're most likely to just postpone it or mention it in passing in specs
21:25:05 [DanC_]
that is: I will
21:25:12 [DanC_]
ACTION DanC: consider adding 'privacy/security' issue, even if we're most likely to just postpone it or mention it in passing in specs
21:25:52 [AndyS]
Eric notes that the current spec can be a base to support this style of interaction
21:26:16 [AndyS]
Bijna: there exists many out of band solutions
21:28:01 [AndyS]
Suggestion for Q: March 31
21:30:28 [patH]
I have to ring off, sorry.
21:30:43 [Zakim]
21:33:11 [AndyS]
Kendall notes 2 pending issues for QL: fromUnionQuery and syntax
21:33:39 [timbl]
21:33:50 [timbl]
21:34:51 [AndyS]
a QL editor on vacation April 1 - 8
21:37:14 [AndyS]
Maybe LC candidate March 31
21:39:49 [DanC_]
... for QL
21:39:53 [AndyS]
Maybe LC candidate for the QL March 31
21:40:11 [timbl]
s/background graph/knowledge base/g
21:42:04 [DanC_]
DanC: anybody up for building one protocol test case?
21:42:15 [DanC_]
SH: yeah^H^H^H... er... maybe
21:43:52 [DanC_]
Topic: Protocol testing
21:44:35 [DanC_]
KC: maybe 4 Apr for LC candidate protocol.
21:46:08 [AndyS]
DanC: hope to close syntax issue March 8 telecon
21:46:24 [AndyS]
.. at a telecon chaired by KC
21:47:47 [AndyS]
Eric would need to email Wed/thurs (full Friday and Monday)
21:47:56 [AndyS]
Eric would need to email Wed/thurs (full Friday and Monday to read by WG)
21:48:02 [DanC_]
Zakim, close agendum 6
21:48:02 [Zakim]
agendum 6 closed
21:48:03 [Zakim]
I see 3 items remaining on the agenda; the next one is
21:48:05 [Zakim]
2. issue: fromUnionQuery [from DanC_]
21:48:16 [DanC_]
Zakim, close agendum 2
21:48:16 [Zakim]
agendum 2 closed
21:48:17 [Zakim]
I see 2 items remaining on the agenda; the next one is
21:48:18 [Zakim]
7. update/insert/manage [from DanC_]
21:48:22 [DanC_]
Zakim, next agendum
21:48:22 [Zakim]
agendum 7. "update/insert/manage" taken up [from DanC_]
21:49:01 [Zakim]
disconnecting the lone participant, OlympiaC, in Team_(dawg)20:44Z
21:49:02 [Zakim]
Team_(dawg)20:44Z has ended
21:49:06 [Zakim]
Attendees were OlympiaC, PatH
21:52:07 [AndyS]
DanC: update on edge of charter
21:53:21 [AndyS]
DanC: soonest to do it is now w/ CG discussion but many inputs
21:53:44 [AndyS]
.. story telling .. scoping
21:54:31 [AndyS]
... reseach:
21:57:11 [AndyS]
Bijan: prior work on ontology evolution
21:57:20 [kendall]
Diff format for RDF graphs and OWL ontologies using Annotea
21:57:50 [ericP]
BijanP: If not updating the expressed graph, deleting is difficult
21:57:55 [kendall]
Biggest issue: if yr not updating the told graph only, then you have choices about deletion... Deleting an inferred triple might require truth maintenance.
21:58:15 [ericP]
... worse than view update problem
21:58:34 [ericP]
... could assemble literature survey for related work on RDFS
21:58:46 [kendall]
pretty straightforward with told graphs; gets really hard in the presence of inference
21:59:12 [kendall]
Being able to add/delete told triples would be of near term use to UMD clients
21:59:56 [ericP]
... currently always PUT documents
22:00:08 [kendall]
Lifesci guys @ IBM...
22:00:15 [kendall]
Only working w/ told graphs right now
22:00:23 [ericP]
Alias: all our apps need RW
22:00:27 [kendall]
but everything is collaborative, so they need to be able to ready *and* write
22:00:35 [kendall]
Elias, eripc :>
22:00:40 [ericP]
22:00:52 [kendall]
V. interested in change history over graphs
22:00:53 [ericP]
... same triple can live in the store twice os we use reification
22:01:01 [kendall]
for auditability
22:01:25 [ericP]
... packaging changes (transactions)
22:01:37 [kendall]
interested in transactions for graph changes; and managing resource contention
22:01:43 [AndyS]
transactions; region locking
22:01:49 [ericP]
... locking, update/merge before write
22:02:04 [kendall]
notification of graph changes by giving server a pattern -- via JMS
22:02:08 [AndyS]
22:02:15 [AndyS]
22:02:17 [ericP]
... notification system for graphs
22:02:18 [AndyS]
22:02:55 [kendall]
Elias: have you seen ARRESTED... REST++. It's Rohit Khare's dissertation. Adds asynch notification to "the web"
22:03:13 [ericP]
... do we use XML documents
22:03:24 [ericP]
[for atomicity of changes]
22:05:25 [ericP]
... with reification, we've be able to address [update/delete] individual statements
22:05:36 [ericP]
SteveH: done some work on diffs
22:05:45 [ericP]
[missed steve's first point]
22:06:08 [SteveH]
swh: Have been playing with SQL-style INSERT and UPDATE
22:06:23 [ericP]
Sato: welcome danc's proposal from a user's perspective
22:06:31 [SteveH]
swh: but deployed code is just POST/PUT to replace or augment an exisitng graph
22:06:34 [Yoshio]
Is it necessary to delete/make others delete obsolute triples? just asserting them as obsolute is not enough? (leaving the others decide whom/what to believe)
22:06:58 [ericP]
... worried about user experience if there is inconsistency consequence of an update
22:07:30 [ericP]
YoshioF: just asserting statements is enough. leave for user to decide what it correct.
22:07:41 [ericP]
... inconsistency could be nature of RDF world
22:08:07 [kendall]
22:08:49 [ericP]
Jordi: so, i hvae a gene ontology and only update. up to app to notice that V3 is more relevent than V2
22:09:11 [ericP]
TimBl: subscription to a form of query is a dream
22:09:50 [kendall]
it means a "true contradiction"
22:09:51 [ericP]
TimBl: interested in applying a patch derived from graph to another
22:10:03 [bijan]
also google for paraconsistent logic
22:10:15 [kendall]
formal study of contradiction (sorta)
22:11:02 [ericP]
DanC: BrianT's customers are gangbusters on update
22:12:03 [ericP]
KendallC: W against told triples is worth standardizing
22:13:21 [ericP]
AndyS: groupware problems different from RSS diffs. solving all is like boiling the ocean
22:13:34 [ericP]
... can we pick a subspace for a charter?
22:14:43 [ericP]
JosD: we cope wiht changes on different levels
22:14:53 [patH]
hey, just tuned in. RDF world, strictly, has NO inconsistency. OWl world, now, another matter.
22:14:57 [ericP]
... clinical practice guidelines change slowly
22:15:24 [kendall]
ericp is dying :>
22:15:41 [ericP]
suppressing a sneeze lead to a big pop from my sternum
22:15:56 [patH]
Kendall, lie him down and sit on him.
22:16:13 [ericP]
JoseD: additive behavior, taking snapshots
22:16:14 [timbl]
q+ to note in response to Andy that in limited experience, though in different applications/situations the algo for generating diffs and the properties of the diffs vary, the wire format for a patch and the patch algo seemed to be constant.
22:16:18 [kendall]
path: that's what was killing him! I've been enjoying my desserts lately!
22:16:42 [kendall]
JacekK: do we want to do any kind of graph update thing?
22:17:06 [patH]
OK, Yoshio, I understand. BUt SPARQL is aimed as RDF-specific QL, right? And RDF itself has no inconsistency.
22:17:18 [ericP]
TimBl: lots of different apps and different diff alogorithms, but the diff format was always the same
22:17:50 [timbl]
22:17:54 [timbl]
22:18:03 [kendall]
we want something that seems a lot more primitive (?) than diffing and patching graphs and sending those around. But -shrug-
22:18:36 [Yoshio]
patH: so my guess is the current issue is something to be treated elswhere than SPARQL deals...
22:18:57 [patH]
Yes, i wuld hope so. We ahve anough trouble with ===.
22:19:34 [AndyS]
It's an informing open discussion
22:19:58 [patH]
OK, didnt mean to rain on parade.
22:20:26 [ericP]
KendallC: want a method in the protocol to send triples to a graph. also want delete (results of a query)
22:20:43 [patH]
While being open, though, I suggest worrrying too much about consistency isnt useful, eg dont thik we need paraconsistency to handle it in practice.
22:21:02 [ericP]
timBl: sounds like what we are doing
22:21:05 [JacekK]
22:21:07 [ericP]
JoseD: can you delete rules?
22:21:26 [ericP]
ack JacekK
22:21:41 [patH]
Can anyone stop you deleting rules?
22:22:07 [patH]
timBl, I'd like to know more about what you guys are doing. POint?
22:22:15 [ericP]
i think it was quesiton of implementation
22:22:44 [ericP]
JacekK: do we want direct change of the graph? or submit somehting to a processor that may or not result in a change.
22:22:48 [ericP]
22:23:10 [kendall]
some people wld argue that much POST usage is mis-usage. :>
22:23:14 [ericP]
... more like POST.
22:23:14 [timbl]
22:23:25 [timbl]
q +ericp
22:23:31 [timbl]
q+ ericp
22:23:53 [ericP]
Jacek: when people use POST, they defer to the service.
22:24:00 [kendall]
q+ to ask what a separate way of representing a graph diff buys over using sparql queries.
22:24:05 [ericP]
... maybe SPARQL should stay at that level
22:24:29 [ericP]
... leave it up to higher level protocols
22:25:06 [patH]
kendall, imagine trying to track changes in a dynamic graph resource, eg produced from a newsfeed.
22:25:23 [AndyS]
Hmm - update/diff language vs/as well as update services
22:25:48 [kendall]
hmm, i'll chew on that, path
22:26:29 [JacekK]
q+ to respond to "just doing a service that updates a graph" and "full deployment of PUT and DELETE" as opposed to availability
22:26:37 [ericP]
ack eric`
22:26:41 [ericP]
ack ericP
22:26:43 [ericP]
ack kendall
22:26:43 [Zakim]
kendall, you wanted to ask what a separate way of representing a graph diff buys over using sparql queries.
22:27:21 [ericP]
zakim, space for 3?
22:27:22 [Zakim]
ok, ericP; conference Team_(dawg)22:27Z scheduled with code 83261 (TEAM1) for 60 minutes until 2327Z
22:27:33 [ericP]
zakim, please dial OlympiaC
22:27:33 [Zakim]
ok, ericP; the call is being made
22:27:34 [Zakim]
Team_(dawg)22:27Z has now started
22:27:35 [Zakim]
22:27:44 [ericP]
PatH, feel free to dial
22:28:05 [Zakim]
22:28:52 [ericP]
TimBl: updtae is like a query except UPDATE, DELETE or BECOMES instead ofn CONSTRUCT
22:29:21 [AndyS]
+1 to update language appraoch
22:30:20 [ericP]
KendallC: any truth maint?
22:30:37 [timbl]
22:30:45 [ericP]
TimBl: no. need to get functional properties
22:30:51 [ericP]
ack AndyS
22:31:09 [ericP]
AndyS: seem to be in a rules world
22:32:03 [timbl]
22:32:12 [ericP]
JacekK: we are mapping to HTTP GET and update mappings would probably get mapped to PUT, DELETE, POST
22:34:36 [timbl]
22:34:47 [timbl]
22:34:51 [JacekK]
ack me
22:34:51 [Zakim]
JacekK, you wanted to respond to "just doing a service that updates a graph" and "full deployment of PUT and DELETE" as opposed to availability
22:34:51 [timbl]
ack Ja
22:36:21 [DanC_]
Zakim, close this agendum
22:36:21 [Zakim]
agendum 7 closed
22:36:22 [Zakim]
I see 1 item remaining on the agenda:
22:36:23 [Zakim]
8. WSDL mapping input [from DanC_]
22:36:32 [DanC_]
Zakim, next agendum
22:36:32 [Zakim]
agendum 8. "WSDL mapping input" taken up [from DanC_]
22:38:36 [ericP]
BijanP: WSDL is defined in terms of a compent model
22:38:56 [ericP]
... components are related to other components or sets of components
22:39:38 [ericP]
... for wsdl-strait, i'm trying to map as closely as possible to the component model
22:40:28 [ericP]
... using 1:1 correspondence with very similar names
22:49:31 [ericP]
[discussion of validation with Bijan's scheme]
22:50:19 [ericP]
... the validation will happen on the WSDL documnets, not on the RDF
22:55:04 [ericP]
... How would this affect your use of WSDL?
23:02:03 [patH]
I gather that Im missing some really terrible thngs on a whiteboard.
23:02:27 [DanC_]
Bijan projected 2 WDSDL-mapped-to-OWL things, yes
23:04:24 [DanC_]
23:04:25 [Zakim]
23:04:41 [Zakim]
23:04:42 [Zakim]
Team_(dawg)22:27Z has ended
23:04:44 [Zakim]
Attendees were OlympiaC, PatH
23:15:52 [patH]
23:15:56 [patH]
patH has left #dawg
23:22:24 [DanC_]
DanC_ has joined #dawg