IRC log of dawg on 2004-09-16

Timestamps are in UTC.

08:11:57 [RRSAgent]
RRSAgent has joined #dawg
08:18:37 [DanC_lap]
Zakim this is DAWG
08:18:41 [DanC_lap]
Zakim, this is DAWG
08:18:41 [Zakim]
sorry, DanC_lap, I do not see a conference named 'DAWG' in progress or scheduled at this time
08:18:48 [DanC_lap]
08:18:50 [DanC_lap]
08:20:05 [kendall]
In attendance: DanC, EricP, SteveH, Yoshio, Jean-FrancoisB, KendallC, JanneS, JosDeRoo, DaveB, AndyS
08:21:25 [kendall]
Also: Dirk-W.vanGulik, AlbertoR
08:22:34 [DanC_lap]
agenda requests: tests? implementations? should show up in due course...
08:24:07 [DanC_lap]
--- naming, marketing, intro, abstract
08:24:14 [DanC_lap]
DaveB: Data Access for RDF: Query. DARQ
08:24:22 [kendall]
marketing talk
08:24:36 [DanC_lap]
(there are tools that will help if we do...)
08:24:36 [kendall]
abstract needs to be a seller, since it's the only thing many people will ever see
08:24:46 [DanC_lap]
Topic: naming, marketing, intro, abstract
08:25:14 [kendall]
googlemark possible for "DARQ"
08:27:09 [kendall]
dirk worries that not having "ql" means it's a bit harder to market corporately
08:27:26 [DanC_lap]
DaveB: DARQL could work
08:27:53 [DaveB]
pronounced "darkle"@en-uk
08:28:15 [kendall]
too many name changes could be problematic
08:30:55 [AndyS]
08:31:09 [AndyS]
08:31:17 [kendall]
we're gonna do a naming poll, apparently
08:31:24 [DaveB]
-1 to RQL as too generic for me
08:31:40 [DanC_lap]
ACTION EricP: set up a naming poll
08:31:47 [AlbertoR]
AlbertoR has joined #dawg
08:33:56 [DanC_lap]
DanC: buzzwords: aggregation, distributed systems
08:34:12 [DanC_lap]
Kendall: query language for the Semantic Web. Dirk: yeag
08:34:14 [DanC_lap]
08:36:00 [JFBaget]
JFBaget has joined #dawg
08:37:12 [DanC_lap]
Kendall: if we could take the intro of the use cases document and turn it around to be positive... that could work
08:38:51 [kendall]
Janne suggests that we ape the way SQL is pitched
08:39:06 [JanneS]
08:40:07 [DanC_lap]
DaveB: ... easy access to data... for programmers
08:40:26 [DaveB]
s/to data/to rdf data/
08:40:33 [kendall]
ACTION KendallC: work on an abstract for the spec
08:41:55 [kendall]
Let's talk DAWG services...
08:42:11 [kendall]
+2 from KendallC
08:43:24 [kendall]
DanC shows off some ideas for DAWG-powered services
08:43:30 [kendall]
TAG last call comments thingie
08:45:10 [DaveB]
08:47:58 [kendall]
many interesting data access services won't do arbitrary graph matches in DanC's opinion
08:48:08 [ericP]
GET -H "Accept: application/rdf+xml"
08:48:33 [kendall]
the protocol design should (eventually -- no, sooner :>) be able to tell clients what sorts of queries it can handle
08:48:50 [kendall]
which gets several +1s from members
08:49:30 [dirkx]
dirkx has joined #dawg
08:49:34 [AlbertoR]
curl --header "Accept: application/rdf+xml"
08:51:18 [ericP]
for N3, GET -H "Accept: text/x-n3"
08:52:08 [DaveB]
DanC - annotea, link to blog world somewhat?
08:52:16 [kendall]
more service ideas: get annotea to do trackbacks using dawg queries, powering the media access use case, for example
08:52:28 [ericP]
annotea server
08:53:46 [ericP]
one with a DARQ input field:
08:53:47 [DaveB]
kendall we have walking tours of the District of Columbia (USA)
08:53:59 [ericP]
one with a DARQ input field:
08:54:00 [DaveB]
.. would like to have clients be able to submit queries to that data
08:54:25 [kendall]
creation commons search engine
08:55:14 [DaveB]
RobS arrives
08:55:17 [kendall]
ACTION KendallC: talk to creative commons people about this stuff
08:56:07 [DaveB]
alberto - newsblender demo
08:56:13 [DaveB]
(offline today)
08:56:49 [DaveB]
searching news as calendar
08:57:04 [kendall]
ACTION KendallC: expose our walking tour data to darq querying clients
08:57:08 [DaveB]
multiple stores
08:57:09 [AlbertoR]
08:57:50 [AndyS]
08:57:52 [AlbertoR] (offline at the moment can return RDF/XML, iCal, N-Triples)
08:58:32 [DaveB]
braces and brackets
08:58:40 [DaveB]
braces={} brackets=() ?
08:59:24 [kendall]
SteveH has 15M triples of CS (??) stuff that will be darq'able at some point
08:59:39 [DaveB]
I can add brql to in a few minutes
09:00:17 [dirkx]
DaveB: My worry is to make sure that we also have negative test cases; as otherwise vendors (like us) will essentially do what big customers expect; and make things as close to what their 100.000 ODBC programmers expect. So in not time you'll be forced to accept (' in place of '{' to get the monkeys of your back..
09:00:28 [dirkx]
So do have negative test cases if this is what we really want!
09:00:54 [DanC_lap]
DanC: I'm going to XML 2004. could present "these things are RDF services from DAWG participants... will interoperate soon (if not already)"
09:01:20 [SteveH]
UK computer science data: - and UI:
09:01:34 [kendall]
SteveH: thanks
09:02:09 [kendall]
Snarky comment of the day, candidate 1: "H1, learn to use H1"
09:06:00 [AndyS]
09:06:10 [AndyS]
That's a describe-like query
09:06:25 [AndyS]
It gets the object (book) and the author details
09:06:37 [kendall]
DaveB: a part of the foaf world can go up as a public demo
09:08:05 [AlbertoR]
09:10:01 [AlbertoR]
09:11:15 [ericP]
LiSci demo
09:13:01 [kendall]
Other presentation venues: ISWC 04, W3 LifeSci workshop, XML 04, publish something on an O'Reilly site
09:13:13 [DaveB]
09:13:18 [DaveB]
deadline for papers, yesterday :)
09:13:43 [AlbertoR]
RDFLets: simple RDF based template system which takes as input a RDQL (BRQL?) query, RDF data source and XSLT - the XSLT is applied over the RDF-for-XML ( result of the query over the RDF data.
09:15:49 [AlbertoR]
E.g. ( which is behind the page) - see query and template - data is FROM RDF/XML URL
09:16:06 [libby]
libby has joined #dawg
09:16:54 [kendall]
The BRQL spec review portion of our show begins now...
09:18:13 [kendall]
Talking now about structure of test cases
09:18:55 [DaveB]
AndyS shows
09:22:19 [ericP]
andy, is actively following the changes to the schema?
09:22:27 [DaveB]
09:22:30 [DanC_lap]
let's look at "A few simple tests"
09:23:52 [DanC_lap]
DaveB: ... problem... I had to have uri-scraping knowledge... which is just wrong
09:25:03 [AlbertoR]
09:25:25 [AlbertoR]
(see Classes - RDF/XML, N-Triples and so on)
09:28:16 [kendall]
Test stuff needs additional work
09:28:54 [Yoshio]
re:comparison but, it is not that easy when bnodes are involved, I guess
09:29:01 [AlbertoR]
still points out previoud work already done for SquishQL/RDQL testing - (thanks to Libby/DanBri for summarising)
09:29:24 [AlbertoR]
09:33:19 [Yoshio]
* N3 as vulgar Latin compared to the holy RDF/XML?
09:33:21 [AndyS]
09:36:02 [DanC_lap]
ack andys
09:38:01 [JosD]
09:41:00 [kendall]
(back on test case stuff: will using the existing xml result set format, which is admittedly not optimized, tie us to the unoptimized version much?)
09:47:41 [kendall]
Conversation about variable prefix choices:
09:47:44 [kendall]
09:47:45 [kendall]
09:47:47 [kendall]
or both?
09:48:09 [kendall]
Main motivation for $foo seems to be to avoid massively installed base of (admittedly) buggy software for stuff like jdbc/odbc (?)
09:48:21 [kendall]
Main motivation for ?foo seems to be momentum
09:48:22 [ericP]
just to be goofy, i added "| 'ñ' Identifier" and it worked
09:48:40 [kendall]
(I hope that's a fair representation, since I sort of care about this.)
09:48:49 [DaveB]
existing rdf query (table, select) results can be and are returned through DBI, *ODBC
09:49:05 [Yoshio]
ericP? I can't read your expression
09:49:56 [AlbertoR]
$foo would allow right the way to ship DAWG/BRQL queries over *existing* DBC interfaces and already deploied software e.g. run DAWG queries inside a M$ Excel spreadsheet
09:49:57 [ericP]
yoshio, you probably need to tell your irc client to use utf-8. i used the ever-popular enya symbol to indicate variables
09:50:11 [Yoshio]
I'm sorry, but I can't
09:50:38 [rob]
rob has joined #dawg
09:50:49 [rob_]
rob_ has joined #dawg
10:00:25 [Yoshio]
I didn't know she has a famous symbol, like Prince. : )
10:02:13 [kendall]
Hmm, it's been pointed out that my summary -- "buggy software" -- isn't particularly accurate. Oh well, ignore at yr leisure.
10:27:49 [dirkx]
AlbertoR: not just DBC; also consider things like IDE's (some of them give you an error when you type somehting which looks like SQL but aint) OR code where the java, c or c++ is pre-prossed by some sql processor - and having both SQL and Java makes it a pain.
10:29:17 [dirkx_]
dirkx_ has joined #dawg
10:29:31 [DanC_lap]
--- resume from break
10:29:41 [DanC_lap]
EricP: I'm ambivalent on ? or $ [or both?]
10:30:00 [DanC_lap]
AndyS: I see some expectation from our semweb audience for ?foo. on the wire syntax matters
10:30:12 [kendall]
Jos agrees re: ?
10:30:15 [kendall]
as do I
10:30:22 [rob_]
Is it worth going for consenus on just having one and not both?
10:30:23 [DanC_lap]
EricP: not both
10:31:11 [DaveB]
DanC let the editor's consider it
10:31:25 [DanC_lap]
ACTION DirkX: provide details about DB interfaces, re ? and $
10:31:52 [dirkx_]
Dirk Will send list with usual suspects which are troublesome; postgress, oracle, dbi, jdbc, odbc and the connections to java, perl, c++
10:32:20 [DaveB]
10:33:41 [Yoshio]
q+ to ask (re 2.3) if one can join triples with ";" like in N3?
10:38:02 [DaveB]
missing . at end of where { triples . }
10:38:09 [DaveB]
is optional, ISSUE
10:38:13 [ericP]
10:38:28 [ericP]
let's just use the agenda for these issues
10:38:38 [ericP]
agenda + missing . at end of where { triples . }
10:41:05 [DaveB]
pls add defn "RDF term"
10:42:44 [DaveB]
10:43:15 [ericP] introduces Propositions and Assertions
10:44:04 [DaveB]
people like accepting RDF term
10:44:10 [DaveB]
10:44:17 [ericP]
kendall, specifically
10:45:18 [DaveB]
JF - why distinction between query variable and rdf variables?
10:45:31 [kendall]
er, i will accept it as written, now. :>
10:45:39 [DaveB]
AndyS - query variables have different range (graph)
10:46:50 [DaveB]
if you want to make mathematical defns, please help out
10:46:56 [DaveB]
by sending some text to the editors
10:49:04 [DaveB]
Yoshio: a query variable is not a name
10:49:15 [DaveB]
DanC: the defn of query variable does not constrain anythihng
10:49:33 [kendall]
AndyS: "An RDF Term is a member of the set formed by the union of U, L, V, and B."
10:49:34 [DaveB]
discussion of definitions standing alone, or with forward referecnes
10:50:01 [DaveB]
for myself, RDF Term is one of (RDF URI Reference, RDF Literal, RDF bnode or Query Variable)
10:50:34 [kendall]
DaveB: yes, that's what my language is supposed to mean. I agree.
10:50:46 [dirkx_]
"A query variable is a name assinged to a graph"
10:54:33 [DaveB]
Alberto - noted B used for bnods and bindings
10:54:40 [DaveB]
andys changed it to BNODES
10:54:53 [DaveB]
AndyS - mapping for bindings includes allowing variables binding to variables
10:55:23 [DaveB]
VAR(b) and VAL(b)
10:56:18 [AlbertoR]
Alberto: noted that the definition of binding B is a function from V to A - but A contains V already - wondered if correct
10:57:06 [JFBaget]
JFBaget has joined #dawg
10:58:08 [DaveB]
AndyS changes RDF Terms to be just Union (RDF URi referneces, literals, bnodes)
10:58:39 [DaveB]
questiion about Var(b) what is b?
11:01:03 [DaveB]
discussion of definition of Binding
11:01:48 [DaveB]
RobS reminds us about binding and binding sets previously discussed
11:02:20 [DaveB]
"set of bindings"
11:02:23 [DaveB]
has precedence
11:02:53 [kendall]
in UC&R 3.2
11:03:12 [DaveB]
AndyS defines 'Binding' as a pair
11:03:37 [rob_]
A binding is a pair x,y where x is a variable and y is an RDF term.
11:04:12 [DaveB]
DanC suggests "A substitution" as the name for a "set of bindings"
11:04:51 [kendall]
Definition: A substitution is a finite set (possibly empty) of pairs of the form Xi=ti, where Xi is a variable and ti is a term, and Xi\neq Xj for every i \neq j, and Xi does not occur in tj, for any i and j.
11:05:02 [kendall]
-- from
11:06:04 [DaveB]
DanC - trying to introduce by example and have the math
11:06:08 [DaveB]
11:07:08 [DaveB]
get sec 2 done tonight & tests for review tomorrow?
11:07:26 [kendall]
ACTION AndyS: fix def'ns in sect 2
11:07:44 [DaveB]
... up to 2.2
11:10:48 [DaveB]
tests discussion
11:10:48 [JanneS]
Zakim, to ask about binding structure in detail
11:10:48 [Zakim]
I don't understand 'to ask about binding structure in detail', JanneS
11:10:49 [SteveH]
q+ to talk about BRQL syntax
11:10:54 [DaveB]
11:11:00 [DaveB]
editing data, results, query
11:11:08 [DaveB]
if agreeing to test, agree to all the syntax
11:11:16 [JanneS]
q+ to ask about binding structure
11:11:57 [DaveB]
kendall concerned about the xml result binding
11:12:08 [DaveB]
in the README?
11:12:31 [DaveB]
other document
11:12:46 [DanC_lap]
ack yoshio
11:12:46 [Zakim]
Yoshio, you wanted to ask (re 2.3) if one can join triples with ";" like in N3?
11:12:56 [DaveB]
Yoshio 2.3
11:13:16 [DaveB]
n3 manner lines seperated by .
11:13:47 [DaveB]
would like to see that in there
11:13:57 [DaveB]
notational convienence, less chars to write
11:13:59 [DaveB]
like N3
11:14:09 [DaveB]
11:14:13 [kendall]
(to follow up: the xml language for returning result sets is not part of the test mechanism)
11:14:14 [DaveB]
... peopel expect that
11:15:02 [DaveB]
Yoshi, wanted to add a ;
11:15:07 [DaveB]
as used in N3
11:15:14 [DaveB]
new syntax
11:15:31 [ericP]
q+ to bring up bnodes
11:15:54 [DaveB]
as well as ,
11:16:15 [DaveB]
Yoshi supports ;, .
11:16:49 [DaveB]
straw poll re ; summary 5ish for, 4ish against
11:17:06 [DaveB]
alberto, don't like the syntax at all
11:17:18 [DaveB]
the n3 bits
11:17:49 [DaveB]
robS doesn't like n3, not an xml syntax
11:18:11 [rob_]
not that i don't like it; i think it sends a very bad message about interoperability
11:18:26 [DaveB]
11:18:38 [DaveB]
EricP - people think it's n3, can do other stuff
11:18:43 [DaveB]
might think that
11:18:56 [DaveB]
bnodes might also be expected
11:19:23 [DaveB]
info - rdql doesn't allow bnodes in queries, sort of
11:19:56 [DaveB]
straw poll on , - 4ish for, 5ish against
11:20:49 [DaveB]
11:20:54 [AlbertoR]
11:21:15 [DaveB]
stay tuned for editor's response
11:21:20 [DanC_lap]
ack steve
11:21:20 [Zakim]
SteveH, you wanted to talk about BRQL syntax
11:21:26 [DanC_lap]
q+ yoshio
11:21:33 [DaveB]
SteveH - round brackets liked
11:21:44 [AlbertoR]
11:21:52 [DaveB]
and don't think the {}s like n3 work well, too much like n3
11:22:13 [DaveB]
parenthese () around each triple - liked
11:22:21 [DaveB]
(s p o ) (s2 p2 o2)
11:22:31 [DaveB]
RobS also agreed
11:22:59 [DaveB]
and no outer braces
11:23:11 [DaveB]
(SteveH) but don't mind them
11:23:45 [DaveB]
dirkx - least surprise works here for the (s p o) form
11:23:53 [DaveB]
SteveH - rdql users use this, familioar
11:24:13 [DaveB]
AlbertoS - existing rdql impls have converged on syntax mostly like this
11:24:35 [DaveB]
AlbertoS supports no n3 syntax, round ()s for nesting and don't use .
11:24:42 [DaveB]
usign AND for constraings
11:24:58 [DaveB]
where (s p o ) AND ... constriants
11:25:24 [kendall]
there's been rather a lot, from my p.o.v.
11:25:56 [DaveB]
AlbertoS - how people mostly write queries today is RDQL, such as on jena lists
11:26:04 [DanC_lap]
(is "the jena mailing list" unambiguous?)
11:26:09 [DaveB]
11:26:12 [DaveB]
jena-dev I assume
11:26:36 [DanC_lap]
are there other jena lists?
11:26:39 [DaveB]
AlbertoS - n3 syntax - ; , . {}s and bits
11:26:52 [DaveB]
yes, there's a jena list on sourceforge, mostly for the developers
11:27:24 [ericP]
11:27:26 [DaveB]
JosD - might conflict with some potential list syntax (not presently in the query language)
11:27:29 [AlbertoR]
PREFIX first can be fine - but the syntax should be RDQL like PREFIX foo FOR <> and PREFIX FOR <http://default/>
11:27:31 [DanC_lap]
alberto, which jena lists did you have in mind?
11:27:46 [AlbertoR]
11:28:09 [DaveB]
AndyS - heard from people that there was no diff between constraints and triple patterns, didn't see why they were separate syntaxes
11:28:20 [DaveB]
... 2) nesting and applied to diff graphs (see tucana0
11:28:28 [DaveB]
... 3) seen no complaints on list
11:28:42 [Yoshio]
q+ to propose use of []'s for bnodes... expecting few supports(?)
11:29:57 [ericP]
all those ()s are NOISY
11:29:58 [DaveB]
are we using the queue or should I jump ion?
11:30:23 [DanC_lap]
Zakim, close the queue
11:30:23 [Zakim]
ok, DanC_lap, the speaker queue is closed
11:30:26 [DanC_lap]
ack janne
11:30:26 [Zakim]
JanneS, you wanted to ask about binding structure
11:30:37 [DaveB]
JanneS binding structures
11:31:05 [DaveB]
looking at ?title = "BRQL tutorial"
11:31:09 [DaveB]
data t yping info
11:31:13 [DaveB]
should be in the test cases
11:31:16 [DaveB]
is this an xsd:string
11:31:22 [DaveB]
AndyS - no it's an RDF literal
11:31:28 [DaveB]
would like datatyping info in the client
11:31:43 [DanC_lap]
ack daveb
11:31:50 [DaveB]
should we go to the xml lang issue
11:32:00 [DaveB]
AndyS it's an rdf literal, so has it
11:32:03 [AlbertoR]
ack Alberto
11:33:12 [DaveB]
RoBS suggests adding a test example with an interger result, a decimal example
11:33:14 [ericP]
RobS: adding and example that had a number would make it clear that literals have data types
11:33:22 [dirkx_]
Example; use 2.499,95 or 1,000.00 :-)
11:33:36 [DaveB]
JanneS all the query results seem to have no datatypes
11:33:45 [DanC_lap]
ack daveb
11:34:35 [DanC_lap]
daveb: query syntax should be regular; no abbreviations ala N3 ; nor ,
11:34:38 [AlbertoR]
DaveB: supports the non N3-alike syntax
11:34:46 [DanC_lap]
ack ericp
11:34:46 [Zakim]
ericP, you wanted to bring up bnodes
11:34:59 [DanC_lap]
ack yoshio
11:34:59 [Zakim]
yoshio, you wanted to propose use of []'s for bnodes... expecting few supports(?)
11:35:22 [DaveB]
prep for lunch, dinner
11:37:42 [DaveB]
lunch till 14:00 UK time
11:37:45 [DaveB]
12:34:24 [DanC_lap]
DanC_lap has joined #dawg
12:40:01 [Yoshio]
Yoshio has joined #dawg
12:48:07 [kendall]
kendall has joined #dawg
12:48:53 [DanC_lap]
RRSAgent, pointer?
12:48:53 [RRSAgent]
12:49:00 [DanC_lap]
RRSAgent, make logs world-access
12:57:30 [AndyS]
AndyS has joined #dawg
13:02:04 [rob]
rob has joined #dawg
13:02:04 [rob_]
rob_ has joined #dawg
13:03:16 [AndyS]
13:04:07 [AndyS]
Plan for the evening : WG + anyone else @Chillis (bottom of Park Street) @8pm
13:04:54 [DanC_lap]
scribe: janne
13:05:19 [DanC_lap]
taking another look at the agenda
13:06:20 [DanC_lap]
RESOLVE to accept 7Sep telcon minutes, ammended to show Farrukh present
13:06:22 [JanneS]
accepted 7-Sep teleconf minutes with amendments from
13:06:32 [AlbertoR]
AlbertoR has joined #dawg
13:07:35 [JanneS]
starting to go through issues in
13:10:25 [JanneS]
background: goal to have 1st WD on of design in September 2004
13:12:27 [JanneS]
by last call we should have all issues resolved (last call target on January 2005)
13:12:55 [kendall]
Re: tomorrow's agenda item about 4.6, I wrote an email:
13:12:56 [kendall]
13:13:05 [kendall]
which I claim satisfies my action:
13:13:12 [kendall]
ACTION Kendall: Pester Aditya about scheme/metascheme query support re: SWOOP
13:13:17 [kendall]
ah, argh :>
13:15:16 [JanneS]
kendall: are we really in a hurry to push this in 1 year instead of 2 years
13:15:36 [JanneS]
DanC: yes, let's try standardize what was already on the market when the group started
13:17:53 [JanneS]
DaveB: w.r.t source feature: implemented it, it works.
13:18:37 [JanneS]
DaveB: still puzzled by bNodes, URIs are ok
13:19:04 [AndyS]
13:19:10 [JanneS]
q+ to how null values are reported
13:19:22 [DanC_lap]
Zakim, open the speaker queue
13:19:22 [Zakim]
I don't understand 'open the speaker queue', DanC_lap
13:19:46 [JanneS]
q+ to how null values are reported
13:20:16 [kendall]
relevant section of the spec is the first section 9
13:20:29 [kendall]
"9 Querying the Origin of Statements"
13:20:32 [JFBaget]
JFBaget has joined #dawg
13:20:54 [JanneS]
DaveB: examples 9.1 and 9.2 from the BRQL V1.55 were implemented
13:20:55 [DanC_lap]
Zakim, open the queue
13:20:55 [Zakim]
ok, DanC_lap, the speaker queue is open
13:20:55 [Yoshio]
* Oops, we have two 9s
13:21:19 [JanneS]
13:21:31 [JanneS]
q+ to how null values are reported
13:21:53 [AndyS]
q+ to comment on quads
13:23:11 [AlbertoR]
13:23:28 [DanC_lap]
q+ to ask if the definitions cover SOURCE
13:23:57 [shellac]
shellac has joined #dawg
13:25:02 [DanC_lap]
q+ to ask for clarification about two 9s and 9.1
13:25:56 [JanneS]
EricP: remembering Tucana folks need further comparison of the source variable (Steve needs this, too)
13:26:04 [dirkx]
dirkx has joined #dawg
13:26:34 [Yoshio]
q+ to ask what is meant by "SOURCE ?s clause is ignored", the subsequent triple is ignored? or it is processed as if it appears without SOURCE ?s ?
13:27:06 [kendall]
It would help *a lot* if "9 \\\\\\\
13:27:08 [kendall]
13:27:09 [kendall]
13:27:41 [JanneS]
SteveH: source ?s (?s dc:creator 'Steve') can be managed with bNodes but trickier with URIs
13:28:11 [DanC_lap]
ack jannes
13:28:11 [Zakim]
JanneS, you wanted to how null values are reported
13:28:54 [SteveH]
SELECT ?creator WHERE SOURCE ?s (:a :b :c), (?s, <dc:creator>, ?creator)
13:30:06 [AlbertoR]
SeveH: +1 - we do exactly the same
13:30:33 [JanneS]
JanneS: what about repositories without source support?
13:30:48 [JanneS]
DaveB: return NULL value (V1.55 suggests this instead of leaving binding out completely)
13:30:52 [AlbertoR]
commas should be optionals - at least in last RDQL spec before this WG work
13:30:53 [DanC_lap]
ack andys
13:30:53 [Zakim]
AndyS, you wanted to comment on quads
13:31:53 [JanneS]
AndyS: implementation experience on source
13:32:11 [JanneS]
scribe needs a link to an example (is there one in the dawg archive... will look myself)
13:34:00 [JanneS]
AndyS: non-deterministic retrieval of source from statements due to toolkit implementations (did I record this correctly?)
13:34:17 [DanC_lap]
(test case for pure quads vs. named containers... dirk says it's straighforward to make one)
13:34:27 [Yoshio]
* FROM is in section 8
13:34:59 [AlbertoR]
13:35:36 [dirkx]
13:35:44 [kendall]
Steve: will you scribe here what you guys do with bnodes & the 'system' graph?
13:35:51 [JanneS]
13:35:52 [kendall]
i'd like that on the record
13:35:54 [SteveH]
kendall, I'l try
13:35:57 [kendall]
13:36:22 [dirkx]
Andy: you can of course ask the system which SOURCES do list a certain triple. But do espect multiple anserws.
13:36:39 [SteveH]
We have a "system graph" that contains the provenance infomation, it holds the bnodes that idetify the graphs
13:36:51 [SteveH]
(should say 'had' the current system uses URIs :)
13:37:00 [AndyS]
dirkx - yes because it is not now an RDF question
13:37:16 [dirkx]
AnndyS: agreed
13:37:34 [kendall]
DanC: hmm?
13:38:46 [JanneS]
AlbertoR: different interpretations of FROM: merged graphs vs. distributed queries
13:38:51 [DanC_lap]
ack danc
13:38:51 [Zakim]
DanC_lap, you wanted to ask if the definitions cover SOURCE and to ask for clarification about two 9s and 9.1
13:38:57 [AlbertoR]
identity mangement UC
13:39:18 [ericP]
q+ to propose leaving space for SOURCE in the grammar and punt it to the next version
13:39:33 [kendall]
Alberto: I intend to put a version of that use case into UC&R the next time I update it -- waiting on the UDDI use case
13:40:09 [AlbertoR]
FROM vs. SOURCE - FROM is more for "4.2 Aggregation Graphs", while/and SOURCE might be more for "4.5 Querying Multiple Sources" - see
13:40:47 [DanC_lap]
ack AlbertoR
13:40:48 [JanneS]
DanC: SOURCE appears not to be complete
13:40:49 [DanC_lap]
ack Yoshio
13:41:10 [ericP]
13:41:13 [ericP]
13:41:19 [DanC_lap]
DanC: does SOURCE design include extended definition of graph pattern matching? DaveB: no, not yet
13:41:24 [ericP]
q- AlbertoR
13:41:30 [ericP]
ack Yoshio
13:41:30 [Zakim]
Yoshio, you wanted to ask what is meant by "SOURCE ?s clause is ignored", the subsequent triple is ignored? or it is processed as if it appears without SOURCE ?s ?
13:41:40 [AlbertoR]
and more about "why bNodes as graph-names" see one of last threads / emails on RDFIG (Phil is one of the early implementors of BRQL it seems) - I am wondering what he feels like about SOURCE vs. FROM
13:42:05 [DanC_lap]
(did phil implement {} or ()? I wonder)
13:42:42 [DanC_lap]
ack ericp
13:42:42 [Zakim]
ericP, you wanted to propose leaving space for SOURCE in the grammar and punt it to the next version
13:42:48 [SteveH]
DanC_lap, neither - he did an abstract syntax
13:43:35 [JanneS]
EricP: implemented source some time ago: worried how it gets implemented: suggestion to leave it out from the next version but leave space in the grammar
13:43:53 [kendall]
13:44:38 [JanneS]
EricP: quads can be implemented as a list of values in the 4th slot - is somewhat heavy way to go
13:48:09 [JanneS]
strawpoll on source indicates yes/no and abstain
13:49:13 [JanneS]
DanC: how about shipping quickly without source and getting source standardized quicker in this way
13:49:26 [JanneS]
Kendall: there are implementations for source already
13:49:44 [JanneS]
DanC: yes but in different ways as it appears, more time would mature the interoperability of implementations
13:50:23 [JosD]
q+ to say that we should do the intersection of different designs
13:50:55 [JanneS]
Kendall: we've been researching the source since the beginning of the WG, how will it get easier in the next 6 months?
13:51:18 [JanneS]
DanC: one technical implementation will dominate over time
13:52:03 [JanneS]
Alberto: let's work more on ~10 test cases for SOURCE and see if interoperability can be reached
13:52:31 [JanneS]
SteveH: interoperability should not be hard to reach
13:52:51 [JanneS]
Jos: let's take the intersection of different approaches
13:54:07 [rob_]
13:54:49 [AndyS]
q+ to explain a test case to test for agreement
13:54:56 [JanneS]
Jos: source is crucial but we don't know the complexities involved w.r.t proofs -> define pragmatic source
13:55:32 [JanneS]
Dirk: thinks technical source different from the 'asserter source' needed in proofs
13:56:57 [JanneS]
EricP: we could reduce the problem space by having a variable for the source which can also be checked for e.g. fixed values (correct me Eric if I got this wrong)
13:57:09 [JanneS]
13:57:15 [DaveB]
Eric P proposed could be changed to require
13:57:15 [DaveB]
SOURCE ?s (s p o)
13:57:15 [DaveB]
SOURCE <uri> (s p o)
13:57:15 [DaveB]
but forbid any other use of ?s in a constraint
13:58:11 [JanneS]
Kendall: minimalistic solution to managing source in the 1st spec version
13:58:43 [JanneS]
Kendall: implementors will give with feedback on its more robust version
13:58:53 [JanneS]
s/give with/come with/
14:00:49 [JanneS]
Rob: aggregation use case with RDF not a representative scenario for querying
14:01:08 [JanneS]
AndyS: we think it is (many others second this)
14:01:21 [Yoshio]
I'm getting convinced it's still premature to include in the spec...
14:01:53 [JanneS]
DaveB: volunteers to suggest text for a minimal design (in support of Eric's suggestion)
14:03:10 [JanneS]
1. source can be just 'returning source'
14:03:25 [JanneS]
2. source can be a) constrained and b) returned
14:04:08 [JanneS]
9.1 = 1
14:05:00 [JanneS]
3. source can be fixed but not returned
14:05:09 [JanneS]
9.1 = 3 in fact
14:09:52 [DaveB]
4. SOURCE ?s and SELECT ?s only
14:10:36 [kendall]
ACTION SteveH: real test cases re: current SOURCE design
14:10:42 [kendall]
ACTION AlbertoR: real test cases re: current SOURCE design
14:12:24 [AlbertoR]
AndyS: shows some interesting SOURCE UC - which is very much similar to the one proposed in 3) at
14:14:07 [JanneS]
ACTION AndyS: make test case available as it distinguishes between two designs of implementing source
14:16:07 [kendall]
AndyS: ask about hardness of getting the src graph in the result set
14:16:21 [kendall]
er, i mean, i want to ask AndyS that question
14:16:49 [ericP]
pictures of quads and formulas -->
14:16:55 [AndyS]
It required no extra machinary
14:17:26 [DaveB]
ACTION DaveB: Update the source section 9, add more formal links, update the examples, try to think about extra constraints as EricP proposed (SOURCE ?s and ?s onlin in SELECT). Look at various ppl's source test cases
14:19:46 [JanneS]
Kendall: has surveyd namespace management in addition to rdf query languages
14:20:06 [JanneS]
Kendall: would prefer declarations of namespace before (strawpoll indicates strong support for before instead of after)
14:23:49 [JanneS]
DanC: almost got into real discussion of issue 2: capability for multiple queries
14:24:34 [JanneS]
SteveH: wants to drop nested optionals
14:25:42 [JanneS]
ACTION SteveH: owns issue 'nested optionals'
14:26:59 [JanneS]
ACTION AlbertoB: owns issue 'DESCRIBE'
14:29:34 [JanneS]
ACTION DanC: add issue 'design of conjunction'
14:30:45 [JanneS]
Kendall: will edit issues list
14:32:25 [JanneS]
Rob: needs more formal description for the result (yes/no) for the ASK feature of V1.55 of BRQL
14:32:35 [JanneS]
Rob: doesn't like the extra keyword AS
14:32:39 [JanneS]
14:34:12 [JanneS]
SELECT WHERE conditions might look ugly
14:35:43 [JanneS]
AndyS: current design with ASK is related to description logics evaluation where the answer might not be any bindings but really yes/no only
14:36:46 [AlbertoR]
14:38:00 [JanneS]
Kendall: DL-KB (sp?) relevant here
14:38:02 [JosD]
14:38:10 [rob_]
14:38:28 [JanneS]
ACTION DanC: owner of issue 'yes or no questions'
14:38:33 [kendall]
DIG DL inference engine protocol
14:42:14 [JanneS]
Rob: ASK is, however, useful as special case because of yes/no answers are given to queries which involve ground triples only
14:42:28 [JanneS]
worksFor(Rob,NI) -> true
14:42:31 [AlbertoR]
ack AlbertoR
14:42:32 [JanneS]
worksFor(Janne,NI) -> false
14:43:50 [AndyS]
14:44:28 [JFBaget]
JFBaget has joined #dawg
14:44:38 [JanneS]
ACTION DanC: add new issue 'eddie number of bindings vs. number of graphs'
14:45:30 [JFBaget]
The result of a query with an empty SELECT is a list (that can be exponential) of empty bindings.
14:45:44 [JFBaget]
The result of ASK is a boolean.
14:45:54 [rob_]
that's a bad design
14:46:25 [JFBaget]
The difference between them is that ASK stops working after finding a first solution.
14:50:06 [ericP]
that works for me
15:06:38 [kendall]
danbri... who invited him?! :>
15:07:46 [ericP]
ericP takes over scribing
15:08:03 [ericP]
discussing the constraints and capabilities workshop
15:08:26 [DaveB]
15:11:12 [ericP]
15:11:24 [ericP]
WS-Policy spec (well sort of)
15:11:57 [ericP]
position paper with useful excerps
15:12:00 [ericP]
15:13:31 [AlbertoR]
AlbertoR has joined #dawg
15:16:10 [ericP]
ACTION Kendall: investigate overlap between RDF query and web service constraints
15:16:18 [DaveB]
above url seems ACLed
15:17:56 [DaveB]
ericP points to
15:20:31 [ericP]
15:20:46 [ericP]
sorry for the above ACL. dunno who made it inaccessible
15:20:50 [ericP]
(or why)
15:21:06 [ericP]
[UDDI use case]
15:21:32 [ericP]
Kendall: i think it's just died.
15:21:42 [ericP]
DanC: will come back if it's important
15:22:35 [ericP]
ACTION Hendler: something related to UDDI
15:22:37 [JosD]
Re previous agendum see rdf topic in WS-desc WG f2f Toronto
15:22:39 [ericP]
15:22:40 [ericP]
ACTION DanC: tell semweb coord that the UDDI action was dropped
15:22:52 [kendall]
Hendler took an action to write a UDDI use case.
15:23:31 [ericP]
[DI use case]
15:23:50 [ericP]
ISSUE: collection or container member accessor
15:24:04 [DaveB]
ref to Bags used for the DI case, as noted by andy
15:24:10 [DaveB]
15:24:48 [ericP]
Dirk: DI use case raises preference expressibility.
15:24:59 [ericP]
DanC: hard to do in closed world?
15:25:18 [ericP]
Dirk: QL can do ordering
15:26:10 [ericP]
Janne: these queries are on the device profile, but should be merged with data on the server.
15:26:14 [dirkx]
Dirk: wanted to say; may be QL - but more likely (like the above wss Policy example) can be done in userland after the query.
15:26:52 [ericP]
15:27:12 [dirkx]
Neet to be cognant that there are very large permutation spaces in those spaces; i.e. having a couple of thousand valid answers baded in the device profile is not uncommon.
15:27:40 [ericP]
cries out for ordering and limiting
15:28:39 [DaveB]
15:28:41 [ericP]
ACTION Steve: send mail about the signage project
15:29:46 [ericP]
ACTION Kendall: add DI use case
15:30:03 [ericP]
ACTION Kendall: whatever
15:30:10 [ericP]
action -22
15:30:13 [kendall]
15:31:31 [ericP]
[Alberto talks about his query/XSLT service]
15:32:06 [ericP]
Alberto: the query return ResultSet format in Andy's doc
15:32:08 [DaveB]
15:32:39 [ericP]
... cgi takes query and XSLT
15:33:54 [JanneS]
q+ to check data typing with this structure
15:34:16 [ericP]
... format is XML so we got to do lots of stuff with XSLT that we couldn't do with the QL
15:35:08 [SteveH]
15:35:16 [AlbertoR]
15:35:30 [ericP]
SteveH: we do something similar. using a different XML result set
15:35:33 [kendall]
Hmm, back on WS constraints and capabilities... Depending on how one represents the state of the world, either in OWL DL or in RDF, you very much could use DAWG to do queries over, for example, service preconditions and effects. It's more likely, as WSDL gets RDFized, that people will be using RDF triple stores to track and represent the states and properties of services, in which case DAWG queries would be very natural to do.
15:35:52 [AlbertoR]
take any query example there and select "RDF Query and Rules Results RDF/XML syntax" output (4th in the radio in the HTML page) - output is XML
15:36:05 [kendall]
And one of our industry partners, Fujitsu Labs of America, which does a lot of this stuff, is itching to use DAWG to support their service composition and planning stuff.
15:36:13 [ericP]
SteveH: our format is terser. that was important.
15:36:35 [JanneS]
agree with Kendall's analysis on the way WS is moving
15:37:11 [AlbertoR] is what I talked about - some very simple preso about this sutff was done at EuroFooCamp in NL last month - and some FOAF demos were shown - see
15:37:58 [macdanbri]
(nice demos :)
15:38:00 [AlbertoR]
to little explain this RDF-for-XML allows a more powerful and flexible Model-View-Controller (MVC) - you give data (RDF), query and XSLT (presentation)
15:38:22 [kendall]
So, it doesn't seem to be the case that it's either simplistic filtering or OWL Full or FOPL.
15:38:26 [ericP]
[discussion of ResultSet and XML datatypes]
15:38:26 [AlbertoR]
btw: we hope to resurrect our main demo machine demo2 soon after disk crash last week)
15:38:54 [ericP]
RobS: most XML processors get their datatypes *only* from schema.
15:39:16 [ericP]
DanC: not xsi:type
15:39:55 [ericP]
[discussion of NULLs]
15:40:24 [JanneS] - '... any design that requires it (xsi:type) is somehow flawed or at least ugly.'
15:42:53 [DaveB]
AndyS says this is used:
15:42:53 [DaveB]
15:42:53 [DaveB]
15:42:53 [DaveB]
<value rdf:datatype="&xsd;decimal">10</uri>
15:42:53 [DaveB]
15:43:16 [ericP]
Alberto: can't we call this a table with rows?
15:43:38 [ericP]
exit dirk, stage left
15:44:12 [ericP]
Kendall: the format should help us document the spec (and visa versa)
15:44:33 [ericP]
... [re SAX-friendly]
15:45:35 [ericP]
... get rid of solution element. doesn't add anything.
15:46:07 [JanneS]
btw, xsi:type for indicating type information needs QName as value, not URI
15:46:27 [ericP]
AndyS: Steve's doesn't have <binding/> element
15:46:59 [ericP]
DanC: neither uses table structure. could use <td/> and not repeat <th/>s
15:49:00 [ericP]
RobS: i'm interested in defining the results. whether it addresses my XQuery access issue is still unknown;
15:49:19 [kendall]
mine looks like:
15:50:14 [ericP]
... binding set, binding element, variable
15:50:21 [kendall]
ej. o
15:50:23 [kendall]
15:50:24 [kendall]
i'll mail it
15:50:39 [kendall]
ACTION KendallC: put xml format up somewhere public, mail a pointer
15:51:12 [DaveB]
looking at
15:55:23 [rob_]
eric: hoping that testing with xquery will demonstrate that it's not a good fit
15:55:35 [ericP]
[XQuery serialization of BRQL]
16:06:38 [ericP]
[investigation of function declaration]
16:06:52 [ericP]
return types optional.
16:07:02 [ericP]
DanC: what is it if it's not declared.
16:07:21 [ericP]
RobS: i documented [in words] the relevent part of the functions
16:07:43 [ericP]
[defining a type for bnodes]
16:08:08 [howardk]
howardk has joined #dawg
16:08:16 [ericP]
heya howard
16:08:22 [howardk]
hi eric!
16:08:45 [howardk]
lo shellac. sorry we couldn't meet in person.
16:08:47 [DaveB]
in 2.6.3 of we find that item / node / * / user-defined type
16:09:02 [DaveB]
so user-defined items can only be sub-types of attributes|document|element nodes
16:09:57 [howardk]
omigosh, you guys are into discussing deep things for so late in the day. :-)
16:10:49 [DaveB]
16:12:07 [ericP]
yeah, it's kinda heavy
16:12:36 [Yoshio]
* too heavy
16:13:30 [DaveB]
looking at
16:13:39 [DaveB]
re returning rdg graphics
16:13:40 [DaveB]
16:19:43 [howardk]
hi macdandri. me too (hey I'm such an irc nubie: how do you do that 'me' thing to make a parenthetical statement??)
16:19:59 [shellac]
/me <some stuff>
16:21:26 [howardk]
is Rob up, trying to convince people of the virtues of XQuery?
16:21:37 [ericP]
RobS: different QLs for every level of the stack will turn people off.
16:24:03 [ericP]
DanC: who is interested in studying XQuery expression?
16:24:13 [ericP]
[straw poll'
16:24:15 [ericP]
16:25:02 [ericP]
RobS, EricP [as path to concensus], andy [interested enough to invest time reading]
16:25:28 [ericP]
/you does
16:30:02 [ericP]
RobS: if this group goes forward with BRQL, i don't know that NI will be able to implement it right away.
16:30:19 [ericP]
Kendall: how about a note?
16:30:31 [AndyS]
Or a member submission
16:30:36 [kendall]
erp :>
16:30:53 [ericP]
RobS: what weight to notes query? who's heard of the OWL XML language?
16:31:16 [kendall]
Kendall: some W3C Notes carry more weight than others. No weight *necessarily*, but no lightness necessarily either.
16:31:25 [kendall]
Depends on what else you and others are willing to put behind it.
16:31:33 [kendall]
and its intrinsic merit
16:50:38 [ericP]
DanC: the note idea has come three or four times. who's interested?
16:50:54 [ericP]
... not so useful to you [RobS]?
16:51:25 [ericP]
RobS: i think it's more sensible for NI to start from XQuery and work from there.
16:52:28 [ericP]
AndyS: we did not discuss using RDF query in non-XML environments
16:52:55 [ericP]
DaveB: i'm concearned about the schedule [meeting Jan LC]
16:54:40 [ericP]
EricP: hoped to provide RDF query access to XQuery users. a note someday would be nice.
16:55:06 [ericP]
Jos: we can solve our problem with SELECT and WHERE. we don't need a full-blown XQuery.
16:56:14 [ericP]
zakim, please pick a scribe
16:56:14 [Zakim]
sorry, ericP, I don't know what conference this is
16:56:20 [ericP]
zakim, this is DAWG
16:56:20 [Zakim]
sorry, ericP, I do not see a conference named 'DAWG' in progress or scheduled at this time
16:56:28 [ericP]
zakim, go home
16:56:28 [Zakim]
I don't understand 'go home', ericP
16:56:38 [ericP]
morning scribe: RobS
16:56:42 [ericP]
16:57:00 [AlbertoR]
AlbertoR has left #dawg
16:59:00 [TomAdams]
TomAdams has joined #dawg
16:59:32 [danbri_dna]
danbri_dna has joined #dawg
17:02:43 [DaveB]
hi Tom, meeting just ended
17:43:00 [TomAdams]
TomAdams has left #dawg
19:13:30 [Zakim]
Zakim has left #dawg