IRC log of dawg on 2005-12-13

Timestamps are in UTC.

14:30:46 [RRSAgent]
RRSAgent has joined #dawg
14:30:46 [RRSAgent]
logging to
14:30:47 [SteveH]
AndyS, I think so, but i'm labouring under the yoke of osx at the moment
14:30:58 [AndyS_]
I didn't touch typePromotion-decimal-decimal-pass
14:32:24 [DanC]
Zakim, this is DAWG
14:32:24 [Zakim]
ok, DanC; that matches SW_DAWG()9:30AM
14:32:25 [Zakim]
14:32:28 [Zakim]
14:32:33 [Zakim]
14:32:54 [DanC]
Zakim, take up item 1
14:32:54 [Zakim]
agendum 1. "Convene, take roll, review records and agenda" taken up [from DanC]
14:32:58 [DanC]
Zakim, who's on the phone?
14:32:58 [Zakim]
On the phone I see [IBMCambridge], LeeF, Libby_Miller, DanC, ??P8
14:33:04 [AndyS_]
zakim, ??P8 is AndyS_+ericP
14:33:04 [Zakim]
+AndyS_+ericP; got it
14:33:06 [EliasT]
Zakim, IBMCambridge is EliasT
14:33:06 [Zakim]
+EliasT; got it
14:33:23 [ericP]
zakim, ??P8 is HP
14:33:23 [Zakim]
I already had ??P8 as AndyS_+ericP, ericP
14:33:34 [ericP]
HP has AndyS_ , ericP
14:33:44 [Zakim]
14:33:48 [ericP]
zakim, ??P8 is really HP
14:33:48 [Zakim]
+HP; got it
14:33:55 [jeen]
Zakim, ??P10 is JeenB
14:33:55 [Zakim]
+JeenB; got it
14:34:00 [SteveH]
AndyS, the xslt tool on this machine is shot, so no, cant rebuild
14:34:02 [ericP]
zakim, HP has AndyS , EricP
14:34:03 [Zakim]
+AndyS, , EricP; got it
14:34:09 [ericP]
zakim, HP has AndyS EricP
14:34:09 [Zakim]
AndyS was already listed in HP, ericP
14:34:10 [Zakim]
EricP was already listed in HP, ericP
14:34:25 [ericP]
zakim, who is here?
14:34:25 [Zakim]
On the phone I see EliasT, LeeF, Libby_Miller, DanC, HP, JeenB
14:34:26 [Zakim]
HP has AndyS, , EricP
14:34:27 [Zakim]
On IRC I see RRSAgent, EliasT, Zakim, LeeF, SteveH, libby, AndyS_, jeen, AndyS, ericP, DanC
14:34:53 [AndyS_]
I get errors on that test ... checking ...
14:34:57 [Zakim]
14:35:04 [ericP]
zakim, ??P8 is really HP999
14:35:04 [Zakim]
+HP999; got it
14:35:07 [ericP]
zakim, ??P8 is really HP
14:35:07 [Zakim]
+HP; got it
14:35:10 [SteveH]
Zakim, [IPcaller] is SteveH
14:35:10 [Zakim]
+SteveH; got it
14:35:12 [ericP]
zakim, HP has AndyS EricP
14:35:12 [Zakim]
+AndyS, EricP; got it
14:35:22 [AndyS_]
I got one result, not zero. I need to check that.
14:35:23 [AndyS_]
14:36:02 [DanC]
Regrets+ Souripriya Das
14:36:09 [AndyS_]
Not clear : datatype(?oneL + ?twoL)
14:36:22 [AndyS_]
Why not xsd:integer?
14:37:02 [DanC]
-> minutes 6 Dec
14:37:30 [DanC]
RESOLVED to accept minutes 6 Dec
14:37:52 [ericP]
scribe: ericP
14:38:27 [EliasT]
I'll be around
14:38:34 [libby]
I'm away
14:38:35 [SteveH]
il be around
14:38:39 [AndyS_]
but I have no idea why I get one result (need to check at length - deep in datatype promotion code)
14:38:54 [DanC]
Zakim, pick a scribe
14:38:54 [Zakim]
Not knowing who is chairing or who scribed recently, I propose EliasT
14:39:07 [ericP]
next meeting: 20 Dec, scribe: LeeF
14:39:54 [patH]
patH has joined #dawg
14:40:10 [DanC]
agenda + punctuationSyntax (comment from Jeremy)
14:40:11 [ericP]
patH, you dialing in straightaway?
14:40:22 [patH]
14:40:26 [DanC]
Zakim, next item
14:40:26 [Zakim]
agendum 2. "issues#rdfSemantics" taken up [from DanC]
14:40:27 [Zakim]
14:40:46 [DanC]
-> pfps's comment
14:41:12 [ericP]
DanC: any progress on PPS's comment?
14:42:13 [DanC]
PatH: sent it... yes, it works as pfps observes and objects to, but I disagree with his argument
14:42:13 [ericP]
14:42:48 [DanC]
^ not relevant
14:43:01 [ericP][]
14:44:34 [ericP]
->[] PatH's impact of semantics message
14:45:13 [ericP]
PatH: a non-lean graph and its lean equivilent, are logically equal
14:46:02 [ericP]
... differing in BNodes only is graph equivilence
14:46:17 [ericP]
... PPS pasted the definition for graph equivilence in the message
14:46:31 [ericP]
... that may have been a pasto
14:47:17 [ericP]
... PPS asserts that lean and non-lean graphs should give the same answers
14:47:20 [ericP]
... I disagree
14:48:10 [ericP]
... I think you should get equivilent answers, but the fat in the graph may manifest in fat in the results
14:49:02 [ericP]
DanC: we need to change our tests if they use logical equivilence
14:49:23 [ericP]
EricP: I don't think anyone is using logical equivilence
14:49:55 [ericP]
DanC: the only test I have uses entailment both ways, i.e. logical equivilence
14:51:00 [ericP]
Andy: I use graph equivilence
14:51:25 [JosD]
JosD has joined #dawg
14:52:18 [ericP]
ACTION: PatH to get Enrico's feedback on PatH's proposal to stick to the Last Call design
14:53:23 [ericP]
Andy: put Enrico's proposal in a note?
14:53:35 [ericP]
DanC: expected it in rq23
14:53:57 [LeeF]
Do I understand correctly that the current state of affairs here is to add text to the spec. which explains the current LC design using simple entailment vocabulary?
14:54:04 [ericP]
PatH: I don't know if there paragraphs that can be pasted into rq23
14:54:20 [LeeF]
(well, current pending a WG vote on rdfSemantics)
14:54:38 [ericP]
Andy: Bijan is using different types of bNodes, which comes from OWL entailment
14:54:42 [DanC]
s/vote/decision/ ("vote" is a 4 letter word that we avoid in W3C, please)
14:54:50 [LeeF]
My apologies.
14:56:04 [AndyS_]
Notation: <_:foo>??
14:56:12 [ericP]
PatH: I endorse allow queries to identify specific bNodes
14:56:31 [ericP]
... using special URIs is a reasonable, but tackier fallback
14:56:43 [SteveH]
Southampton wants it too
14:58:52 [AndyS_]
"Told bNodes" requires a small tweak to the result set format as well.
14:58:59 [ericP]
EricP: I prefer to put it in SPARQL2 'cause of time to engineer queries, tests and implementation
14:59:25 [ericP]
DanC: find it distasteful
14:59:57 [ericP]
... different from RDF model [semantics]
15:00:29 [ericP]
[discussion of existentials and UNA]
15:00:45 [ericP]
[scribe indulges in scribing points that he wants to hear:]
15:00:59 [ericP]
PatH: it would certainly be a lot simpler without them
15:01:14 [ericP]
[scribe done indulging]
15:02:06 [ericP]
SteveH: I have to put it in my code, standardised or not
15:03:21 [ericP]
15:03:36 [DanC]
proposals on rdfSemantics (a) as LC, with "simple entailment" in the right place (b) with told bNode
15:05:04 [DanC]
ACTION PatH: draft told bNode proposal re rdfSemantics, preferably with test case sketches
15:05:07 [ericP]
EricP: can you adivise where to put "simple entailment" in ?
15:06:25 [DanC]
I think " Definition: Pattern Solution " is the relevant place
15:06:47 [DanC]
or maybe " Definition: Basic Graph Pattern "
15:06:56 [DanC]
[[ A basic graph pattern matches on graph G with solution S if S(GP) is an RDF graph and is subgraph of G. ]]
15:07:54 [DanC]
ACTION PatH: suggest where "simple entailment" should go, re rdfSemantics
15:08:15 [DanC]
Zakim, next item
15:08:15 [Zakim]
agendum 3. "issues#valueTesting" taken up [from DanC]
15:08:34 [ericP]
DanC: expect both proposals in the next day or two
15:08:43 [Zakim]
15:08:43 [DanC]
. ACTION: EricP to mail the WG asking about URIs for functions, re comment 2005-10-26T22:08:10Z from r.newman
15:11:09 [DanC]
EricP: took a different tack; removed sop:
15:11:27 [ericP]
EricP: responded to commenter stating that sop: editorial dingus is gone
15:11:38 [ericP]
15:11:40 [DanC]
section 11
15:11:50 [ericP]
ACTION: EricP to mail the WG asking about URIs for functions, re comment 2005-10-26T22:08:10Z from
15:12:03 [ericP]
15:12:08 [ericP]
action -11
15:12:51 [DanC]
15:13:01 [DanC]
_:a foaf:givenName "Björn".
15:13:01 [DanC]
ASK { _:a foaf:givenName ?gn
15:13:01 [DanC]
FILTER regex(?gn, "^Bjoern$")}
15:13:01 [DanC]
15:14:12 [DanC]
doesn't match, per ericp's proposal
15:14:12 [DanC]
15:14:13 [DanC]
[] dc:title "solitario"@ES .
15:14:13 [DanC]
[] my:supportsLang "FR", "ES".
15:14:13 [DanC]
ASK { ?who dc:title ?title .
15:14:13 [DanC]
?app my:supportsLang ?lang
15:14:15 [DanC]
FILTER ( LANG(?title) = ?lang ) }
15:14:19 [DanC]
15:15:08 [DanC]
true? FILTER "abc" = "abc"^^xsd:string ?
15:15:09 [EliasT]
that's annoying
15:16:20 [DanC]
true: FILTER 1.0e0 = 1 ?
15:17:11 [SteveH]
equal for me
15:19:06 [AndyS_]
arq.expr '"abc" = "abc"^^xsd:string' ==>true
15:22:13 [AndyS_]
It's equal because it is built it like that - one line of code control it
15:23:49 [ericP]
15:23:51 [ericP]
The collation for fn:compare is defined by XPath and identified by This collation allows for string comparison based on code point values. Codepoint string equivilence can be tested with RDF term equivilence.
15:23:55 [ericP]
15:24:36 [jeen]
Zakim, mute me
15:24:36 [Zakim]
JeenB should now be muted
15:24:56 [LeeF]
15:25:12 [AndyS_]
Extension function is better for interop
15:25:25 [DanC]
poll: anybody see need for standardized collation support in sparql? no 'aye's
15:25:45 [ericP]
+ xsd:boolean
15:25:51 [ericP]
. + xsd:boolean
15:26:05 [ericP]
. - integrity checks on XML Schema datatypes
15:26:40 [AndyS_]
Test: datatype (?a = ?b)
15:26:44 [jeen]
zakim, unmute me
15:26:44 [Zakim]
JeenB should no longer be muted
15:27:36 [ericP]
. - re-stated type promotion
15:27:39 [AndyS_]
(?a = ?b) = ?bool
15:28:06 [AndyS_]
(?a = ?b) = "true"^^xsd:boolean
15:29:34 [ericP]
ACTION EricP: add test case for support of xsd:boolean (ala (?a = ?b) = "true"^^xsd:boolean )
15:29:36 [SteveH]
only skimmed it
15:29:50 [AndyS_]
skimmed it
15:32:17 [ericP]
. - xsd:boolean result type text per changes in XPath funciton invocation
15:32:44 [ericP]
. ~ changed defn of EBV erp changes in XPath funciton invocation
15:33:17 [ericP]
. + example for langMatches(?x, "*")
15:33:28 [ericP]
. - regex collation
15:33:34 [DanC]
really? = can never type error? how about: 21 = "XXI"^^my:romanNumeral
15:33:51 [ericP]
. - error text in extensibility
15:35:35 [ericP]
15:37:00 [ericP]
we *could* spell = two different ways
15:38:16 [DanC]
(earlier, I argued that the results set should have had syntax for "and there may be more; I'm not sure")
15:45:08 [DanC]
(digression about iTQL importing datatypes as graphs)
15:47:00 [ericP]
[involved discussion about support for =]
15:47:18 [AndyS_]
There is a test for Roman numerals
15:47:49 [AndyS_]
FILTER ( ?n = 21 ) pass when ?n is "XXI"^^roman:Numeral
15:48:50 [LeeF]
We can't say "if you understand roman:Numeral" in the test suite though.
15:49:23 [AndyS_]
Arg - not in the manifest.n3
15:50:00 [DanC]
EricP: I think that's not accepted
15:51:02 [DanC]
DanC: I hope/wish for lots of test cases re valueTesting
15:51:29 [DanC]
EricP: red/green text going away
15:51:30 [DanC]
15:51:32 [DanC]
Zakim, next item
15:51:32 [Zakim]
agendum 4. "issues#wsdlAbstractProtocol (esp query by reference)" taken up
15:51:41 [ericP]
EricP: this is the last change to view it as a set of (sentient) diffs
15:51:43 [JosD]
given :a :b 21 query SELECT ?X WHERE ?X :b "XXI"^^roman:Numeral answers with :a
15:52:02 [DanC]
Zakim, take up item 5
15:52:02 [Zakim]
agendum 5. "comment: SPARQL Results Format and Unbound Variables" taken up [from DanC]
15:52:24 [jeen]
15:52:29 [DanC]
-> Jeen 13 Dec
15:52:29 [AndyS_]
Good summary from Jeen
15:53:13 [DanC]
Zakim, who's on the phone?
15:53:13 [Zakim]
On the phone I see EliasT, LeeF, Libby_Miller, DanC, HP, JeenB, SteveH, PatH, Jos_De_Roo
15:53:15 [Zakim]
HP has AndyS, EricP
15:53:30 [DanC]
which do you prefer? any you would object to?
15:53:47 [ericP]
[around the table, discussing the proposals, as labeled by in]
15:53:57 [ericP]
EliasT: +a, -c
15:54:18 [ericP]
LeeF: +a
15:54:24 [ericP]
Libby: +a
15:54:52 [ericP]
JeenB: +a, -b
15:55:06 [ericP]
SteveH: +b, +c
15:55:24 [ericP]
JosD: +.5a
15:55:40 [AndyS_]
a > c >> b (b well behind c) - "tolerate" b (no formal objection but its yukky)
15:55:42 [ericP]
AndyS: +a, -.5b
15:55:51 [ericP]
PatH: -c
15:55:58 [ericP]
EricP: +a
15:57:44 [EliasT]
This document is based on the binary query results format as available in
15:57:44 [EliasT]
the Sesame RDF Framework (, release 1.2 and later.
15:57:44 [EliasT]
The relevant Javadoc for Sesame's implementation can be found at
15:57:44 [EliasT]
15:57:56 [ericP]
-> summary of Sesame binary results
15:58:46 [DanC]
-> 4.7 Bandwidth-efficient Protocol
15:58:52 [ericP]
JeenB: parsing time is 10 times faster for binary
15:59:11 [ericP]
PatH: space factor is about 10 as well
15:59:13 [EliasT]
would we support both binary and XML formats?
16:00:11 [DanC]
"it shall allow for at least one result format that does not make excessive use of network bandwidth for a given collection of results"
16:01:15 [ericP]
EricP: expect that choosing XML is 500 times more relevent than picking a particular XML serialization for unbound
16:01:36 [DanC]
PROPOSED: to stick with LC design, and cite informatively a binary results format
16:02:11 [ericP]
PatH: trying to space-optimize XML is like trying to build buildings from better chocolate
16:02:26 [EliasT]
it's a legacy from DaveB
16:03:10 [DanC]
(sorry to go long)
16:03:11 [Zakim]
16:03:12 [LeeF]
16:04:16 [ericP]
JeenB: found XSLT for unmentioned unbounds harder
16:04:50 [EliasT]
Jeen just found it hard.
16:05:11 [AndyS_]
It's easier in XQuery than XSLT
16:06:08 [ericP]
PatH: if it will help, i will withdraw my -c
16:08:50 [ericP]
[EliasT draws parallen between the processing effort of both unmentioned unbounds and unordered results]
16:11:25 [DanC]
q+ to propose to adjourn
16:12:10 [EliasT]
16:12:12 [Zakim]
16:12:14 [Zakim]
16:12:16 [SteveH]
16:12:20 [EliasT]
EliasT has left #dawg
16:12:20 [Zakim]
16:12:24 [Zakim]
16:12:25 [Zakim]
16:15:08 [DanC]
Jeen, it would help me out a lot if you would lead a discussion on the mailing list about a vs. c. Take a position on one of them, if only to draw out the arguments on the other side
16:15:16 [jeen]
will do dan
16:15:28 [jeen]
(or: will try, rather ;))
16:25:06 [jeen]
16:25:24 [jeen]
(the relevant bit is the last template in there, named "matchColumn")
16:29:12 [Zakim]
16:29:14 [Zakim]
16:29:18 [Zakim]
16:29:20 [Zakim]
SW_DAWG()9:30AM has ended
16:29:22 [Zakim]
Attendees were LeeF, Libby_Miller, DanC, EliasT, JeenB, AndyS, , EricP, SteveH, PatH, Jos_De_Roo
18:00:59 [SteveH]
SteveH has left #dawg
18:27:02 [AndyS]
AndyS has left #dawg
18:32:59 [Zakim]
Zakim has left #dawg
18:35:19 [LeeF]
LeeF has left #dawg
19:22:09 [libby]
libby has joined #dawg
19:41:54 [libby]
libby has left #dawg