IRC log of dawg on 2004-07-14

Timestamps are in UTC.

16:28:44 [RRSAgent]
RRSAgent has joined #dawg
16:28:57 [kendall]
close issues primarily by adding test cases to a suite
16:30:05 [kendall]
Simon: we have to be careful about "closing issues" by *adding* new features, which may lead to an overly complex language
16:30:47 [kendall]
test-driven devel is supposed to give us a certain amount of "design courage"
16:31:38 [kendall]
bootstrapping problem with articulating tests *in* the language, except some tests and new features change the language
16:31:49 [kendall]
thus, tests will have to be rewritten, perhaps repeatedly
16:32:13 [kendall]
rewriting can be obsoleting tests
16:33:10 [kendall]
we can start with natural language descriptions of tests; but eventually we'll want an automated, machine-processible test suite, so that other folks can interact with them.
16:33:29 [kendall]
we'll also need protocol tests -- which can be a pain
16:34:45 [kendall]
Steve Harris doing some work keeping up with tests
16:35:24 [kendall]
ericp, yoshio, jos, dan, andy, kendall -- have cvs write access
16:36:11 [kendall]
Issues list is being handled by ericp
16:36:13 [Sebkha]
Sebkha has joined #dawg
16:36:18 [kendall]
but he'd like to have some help
16:38:36 [kendall]
con of test driven development is that you need a separate process or structure that worries about design goodness
16:38:46 [kendall]
danc: before we close an issue, we'll need a test
16:38:56 [JosD]
JosD has joined #dawg
16:39:15 [kendall]
good for regression and unforeseen side-effects
16:39:25 [kendall]
err, test drive devel
16:45:50 [kendall]
(group members say where they are re: initial designs...)
16:47:20 [DanC_jam]
RobS: in sum, I like adopting XQuery to querying RDF. I don't like BRQL.
16:48:10 [DanC_jam]
AndyS: I like BRQL. I think the XQuery-based designs leave too much of the burden on the app developer, and don't straightfowrardly express RDF queries.
16:48:41 [DanC_jam]
Simon: the WHERE clause in RDQL looks abstractly simple, so I like that.
16:48:52 [kendall]
Jos: supports N3QL
16:49:42 [kendall]
Jos also likes rdql/brql, prefers n3ql because it's "all in triples"
16:50:19 [kendall]
Jos -- have brql as a strawman and an n3ql serialization (?)
16:53:10 [DanC_jam]
(hmm... perhaps a rules workshop would be useful to inform this WG)
16:53:24 [kendall]
yoshio -- brql and n3ql
16:53:32 [Zakim]
Zakim has joined #dawg
16:54:43 [kendall]
Hiroyuki -- leaning toward brql
16:54:46 [DanC_jam]
HiroyukiS: I think those familiar with the semantic web will support BRQL, being familiar with RDQL, but for the whole web community, I'm not sure...
16:54:55 [Yoshio]
My concern is that the concrete language shold much in common wiht the possible language to be used in the higher level of the layer cake.
16:54:58 [DanC_jam]
... I think BRQL is a reasonable starting piont.
16:56:04 [Yoshio]
The concrete language should be able to be called by the higher languae, or should be extended to the languge
16:58:42 [kendall]
16:59:25 [DaveB]
DaveB has joined #dawg
17:00:54 [kendall]
howard -- xsrql is beautiful, elegant, easy to implemet! :>
17:02:53 [DaveB]
17:03:26 [DanC_jam]
DanC: the RDQL WHERE clause feel like it has 4 corners around it. Parts of BRQL feel like premature standardization, to me, but I'm reasonably confident that we can work the features into a more regular form, ala iTQL or N3QL
17:04:15 [kendall]
ericp -- worries about a false sense of familiarity among end users for a language with sql-like syntax
17:04:41 [kendall]
ericp -- worried about federation tradeoffs
17:06:42 [kendall]
erip -- still prefers algae; n3ql works fine; unsure of the value of it being a series of statements.
17:07:27 [DanC_jam]
KC: not much to add over what I/we put in the survey form.
17:07:30 [ericP]
kendal: i don't have an XQuery processor lying around
17:07:32 [DanC_jam]
[which isn't public yet, btw]
17:08:23 [ericP]
kendal: convincing real world people to use [dawg-ql]
17:08:41 [ericP]
... even a false-sense of familiarty is helpful
17:09:05 [ericP]
... i'm conservative. [RdfDB] family is the most implemented
17:10:11 [ericP]
... therefor refer to RDQL or BRQL
17:10:13 [kendall]
kendaLL <-- KENDAll <-- two l's please :>
17:10:37 [ericP]
17:12:15 [kendall]
December meeting scheduled for 1-2 Dec in Las Vegas
17:13:36 [kendall]
danc proposes to keep it in LV and put it early Jan
17:14:25 [DanC_jam]
s/proposes/leans toward/ (I'm not offering to host)
17:17:11 [DanC_jam]
RESOLVED: to recind the 1-2 Dec ftf date.
17:17:23 [DanC_jam]
ACTION EricP: find out when/where the W3C tech plenary is in 2005
17:20:21 [DanC_jam]
break for :15
17:38:39 [DanC_jam]
hmm... just realized... we're preparing Howard's presentation 20minutes early... who's planning to call in? Just Dave?
17:45:57 [DanC_jam]
Zakim, this will be dawg
17:45:57 [Zakim]
ok, DanC_jam; I see SW_DAWG(f2f)2:00PM scheduled to start in 15 minutes
17:47:54 [afs]
afs has joined #dawg
17:49:44 [DanC_jam]
17:49:53 [kendall]
Howard's XsRQL Presentation
17:50:15 [DanC_jam]
Zakim, remind us in 20 minutes to think again about how long this presentation should last
17:50:15 [Zakim]
ok, DanC_jam
17:51:29 [DanC_jam]
(Howard is projecting from his laptop. no slides etc.)
17:53:14 [DaveB]
I am planning to dial in
17:53:36 [DanC_jam]
JosD: what data are you querying against? Howard: some vCard data I found when looking at AndyS's stuff
17:54:25 [DanC_jam]
KH: parsing with ARP. built my own triplestore in Java.
17:54:31 [DanC_jam]
HK, that is
17:54:47 [DanC_jam]
HK demonstrates dawg:debug-sticky;
17:54:57 [DanC_jam]
which uses XQuery prolog syntax
17:55:50 [DanC_jam]
HK: the prolog thing is useful; could be added to RDQL or whatnot
17:57:44 [JosD]
JosD has joined #dawg
18:00:28 [DanC_jam]
DanC: is it OK to do "distinct" processing even when it wasn't requested? does this proposal specify the count=8 in the non-distinct case? HK: yes
18:00:49 [DanC_jam]
HK: yes, all correct implementations do count=8
18:02:13 [kendall]
@* -> returns all the predicates
18:02:39 [kendall]
@*/* -> returns the object that is connected to every predicate
18:03:02 [kendall]
xsrql -> path language returns the *last* thing in each path
18:03:59 [kendall]
',' operator concatenates sequences
18:07:13 [DanC_jam]
AndyS: can't have sequences of sequences in the Query data model, right? HK: right.
18:08:26 [Zakim]
SW_DAWG(f2f)2:00PM has now started
18:08:33 [Zakim]
18:08:48 [TomAdams]
TomAdams has joined #dawg
18:09:15 [KevinW]
KevinW has joined #dawg
18:09:15 [DaveB-lap]
DaveB-lap has joined #dawg
18:09:22 [Zakim]
18:09:35 [DanC_jam]
Zakim, who is on the phone?
18:09:35 [Zakim]
On the phone I see Kevin, RobS
18:09:45 [DanC_jam]
daveb? wanna call in?
18:09:53 [DaveB-lap]
yup, will do
18:10:15 [Zakim]
DanC_jam, you asked to be reminded at this time to think again about how long this presentation should last
18:10:50 [Zakim]
18:11:00 [DanC_jam]
Zakim, ??P2 is DaveB
18:11:00 [Zakim]
+DaveB; got it
18:14:42 [kendall]
A sample XsRQL query:
18:14:45 [Zakim]
18:15:01 [kendall]
declare prefix vcard: <>;
18:16:02 [kendall]
* [ @vcard:FN = "John Smith" ]/@vcard:N/*/@vcard:Family
18:17:00 [kendall]
oops, make that:
18:17:14 [kendall]
18:19:53 [DanC_jam]
ref: XsRQL: an XQuery-style Query Language for RDF
18:27:29 [AndyS]
for $libby := *[ @foaf:mbox = "" ]
18:27:29 [AndyS]
18:27:29 [AndyS]
{ $libby, @newFoaf:Name, $libby/@foaf:name/* }
18:28:31 [kendall]
will return ->
18:28:53 [kendall]
(<...libby1>, foaf:name, "Libby", <libby2>, foaf:name, "Libby2")
18:28:59 [kendall]
i.e., list is flattended
18:29:05 [kendall]
er, flattened
18:34:20 [DanC_jam]
Zakim, RobS is temporarily MeetingRoom
18:34:20 [Zakim]
+MeetingRoom; got it
18:34:31 [DaveB-lap]
did somebody write down how long Howard's impl estimate was?
18:34:58 [DanC_jam]
I think he said he's 5 or 6 weeks into a 5 or 6 month implementation effort. for an in-memory architecture.
18:40:30 [kendall]
18:40:41 [Yoshio]
q+ to ask "XML is intrinsically a tree, while RDF is not (mesh, or even loopy) , won't this cause any trouble with path language?"
18:41:17 [DanC_jam]
Zakim, who's on the phone?
18:41:17 [Zakim]
On the phone I see Kevin, MeetingRoom, DaveB, Tom_Adams
18:41:21 [TomAdams]
Remote applause
18:44:48 [TomAdams]
Guys, Andy is very soft, can he come closer to the Mike please.
18:45:31 [DanC_jam]
DanC: N3QL's motivation is integration with rules, diff, etc.
18:46:34 [AndyS]
AndyS wants to separate design (functional expressivity) and syntax (the way its written)
18:46:40 [AndyS]
(AndyS scribes)
18:47:14 [AndyS]
Simon: if we have graph results, are we not doing a (simple) rules language
18:47:35 [AndyS]
DanC: rules => different shaped graphs
18:47:53 [TomAdams]
Re Abstract Syntax: I've had Simon ear bash me about this also, and I like it. We use it in Kowari to allow plugging in of different languages, e.g. iTQL & RDQL.
18:48:40 [kendall]
+1 to andy's separation
18:48:46 [AndyS]
Semweb-cg has discussed a rules workshop but no decisions
18:49:49 [Yoshio]
18:50:06 [AndyS]
AndyS: rules systems can include the feature of results feeding back into the KB
18:50:19 [AndyS]
for them, we aren't doing rules yet
18:51:20 [AndyS]
Rob: users want a clear way to write the things they want to do (example "or")
18:51:42 [Zakim]
18:52:23 [AndyS]
not use idioms to achive their tasks
18:53:46 [DanC_jam]
RobS: universals seem to have few precedents in extant languages. e.g. "someone all of whose girlfriends are attractive" is hard to express
18:54:36 [AndyS]
Rob: trade off of expressing in OWL or in the QL
18:56:39 [Yoshio]
q+ to say if our language has capability to handle premises or graph unioning , it could lead to a smooth connection to the rules
18:56:52 [AndyS]
Link to 4.6 here - should we have a ref to rules here ?
18:58:04 [AndyS]
Enrico gave an example of inferred information
18:58:23 [AndyS]
Rob says this can't be expressed in triples: Jos disagrees
19:00:16 [AndyS]
(Jos has emailed the list:
19:00:58 [AndyS]
Dan floats the idea that we have text about things that are a good idea but we can't do in the WG timeframe
19:02:22 [kendall]
the design objectives section of the current doc doesn't do that
19:02:26 [AndyS]
Query processors may advertise their features ; queries may request features
19:02:30 [TomAdams]
To follow up on general "good ideas", these can be punted off onto the SWBP&D WG...
19:04:00 [AndyS]
Could use DObj for these things (good things we may not do)
19:04:24 [JosD]
paper "Description Logic Programs: Combining Logic Programs with Description Logic"
19:04:58 [howardk]
howardk has joined #dawg
19:05:54 [AndyS]
ACTION Simon+Kendall, elaborate the rel. of rules and "construct" in the UC&R doc
19:06:51 [AndyS]
AndyS recalls that PatH noted the user request for transformation of the RDF in his DQL feedback
19:06:59 [DanC_jam]
ack yoshio
19:06:59 [Zakim]
Yoshio, you wanted to say if our language has capability to handle premises or graph unioning , it could lead to a smooth connection to the rules
19:07:59 [AndyS]
Yoshi notes that N3QL can support premises
19:08:18 [AndyS]
19:10:13 [AndyS]
Yoshio: keep the additional triples in the query , not just the URI of the targets
19:10:46 [AndyS]
Dan leans towards protocol support
19:11:30 [AndyS]
Recursive query (inserting results into te federation of the source)
19:11:51 [Yoshio]
19:13:09 [kendall]
19:13:35 [DanC_jam]
19:13:35 [AndyS]
19:13:43 [AndyS]
in public space
19:16:47 [AndyS]
iTQL has a "GIVEN" clause for passing into a query variable bindings (used internally)
19:17:04 [AndyS]
AndyS notes that ASemantics have a use case for passing in initial bindings
19:17:33 [AndyS]
AndyS: this is not premises - adding triples into the query target spaces then querying
19:17:52 [AndyS]
EricP: Algae is based around this (federation)
19:19:00 [Zakim]
19:19:09 [DaveB-lap]
harsh :)
19:19:12 [TomAdams]
Sorry guys... :(
19:19:37 [TomAdams]
Yes, thanks!
19:19:58 [Zakim]
19:20:08 [AndyS]
ACTION EricP, send ref about Algae motivation
19:20:18 [DanC_jam]
. ACTION: EricP offer implementatino experience (in use case form?) with premises and algae
19:20:59 [AndyS]
Eric was more thinking of initial bindings
19:21:01 [DanC_jam]
ericP: hmm... premises or pre-bindings?
19:21:22 [DanC_jam]
ACTION: EricP offer implementatino experience (in use case form?) with premises and algae
19:21:40 [TomAdams]
My apologies all... Didn't realise that the phone system would do that!
19:21:57 [AndyS]
UCs relavant: traffic, media conglomerate
19:22:24 [kendall]
UC 2.3 and 2.5
19:24:49 [ericP]
19:25:12 [DaveB-lap]
dialing out for now
19:25:17 [Zakim]
19:25:18 [AndyS]
Lunch until 2:00pm and talk about techie things : 10min summaries
19:25:34 [DanC_jam]
lunch until 2pm... the lunch is longish, on the expectation that folks will have techincal discussions that they'll be interested to report from afer lunch.
19:25:58 [TomAdams]
Thanks guys, I'll be signing off.
19:26:27 [Zakim]
19:30:23 [Zakim]
19:30:24 [Zakim]
SW_DAWG(f2f)2:00PM has ended
19:30:25 [Zakim]
Attendees were Kevin, DaveB, Tom_Adams, MeetingRoom
20:41:22 [DanC-AIM]
Massimo, have you looked at xsrql?
20:41:59 [DanC-AIM]
Bummer... I should have had you dial in while it was presented/discussed. But I guess you didn't miss too much.
21:02:46 [DanC_jam]
(resuming after lunch...)
21:02:54 [DanC_jam]
oops... wrong channel...
21:03:48 [AndyS]
Eric reports on what he did on his hols
21:04:17 [AndyS]
Context: query federation / query permis
21:04:57 [AndyS]
Ship a set of bindings to seed the next query does not work if the target does inference that you can't
21:05:33 [AndyS]
Contrast: variable bindins vs asserting the triples that caused those bindings
21:06:11 [AndyS]
Also the item from a list issue
21:07:12 [AndyS]
SimonR reports that Rob has convinced him XQuery is a good idea
21:07:21 [DanC_jam]
s/convinced/nearly convinced/
21:08:43 [AndyS]
-- Move to UC&R doc
21:08:44 [Yoshio]
Yoshio has joined #dawg
21:09:04 [AndyS]
3.6 Optional Match
21:09:18 [HiroyukiS]
HiroyukiS has joined #DAWG
21:11:16 [AndyS]
TBL opposed to optionals ; Dan worries about the time cost ; Eric has been persuaded it is needed
21:11:37 [Sebkha]
Optional match advocacy:
21:11:50 [AndyS]
Rob: Users want formal organisation of data
21:12:08 [ericP]
AndyS's OPTIONAL test cases expressed in BRQL:
21:12:10 [AndyS]
Also: it is not a *requirement* as a priority
21:12:12 [ericP]
(now on the mirrors)
21:12:33 [Sebkha]
The fact that users WANT formally organized data doesn't help them when no such data happens to be available.
21:13:44 [JosD]
JosD has joined #dawg
21:15:22 [AndyS]
Rob illustrates with general disjunction is not currently a requirement but is as useful
21:16:31 [AndyS]
Yoshio: danger of non-std extensions
21:16:40 [AndyS]
if not included
21:17:15 [AndyS]
Proposed wording in doc (Kendall, Andy)
21:18:01 [AndyS]
Vote: for: all except against: 1 (Rob) abstain: 1 (Jos)
21:18:29 [AndyS]
ACCEPT: 3.6 with wording as v1.123
21:19:29 [AndyS]
Move to "3.8 Bookmarkable Queries"
21:20:38 [AndyS]
Does imply that query syntax is self-contained (stateless request)
21:21:59 [AndyS]
And is HTTP not just SOAP
21:22:25 [AndyS]
Strawpoll on move to Objective (most in favour)
21:22:44 [AndyS]
Move to "3.10 Result Limits"
21:23:56 [AndyS]
Rob: it makes sense only with sorting
21:24:02 [AndyS]
Others disagree
21:26:49 [AndyS]
Use case: library system / facteed browser
21:26:59 [AndyS]
Agreement: offset+limit => sort
21:27:21 [AndyS]
Kendall: The mobile phone example
21:31:38 [DanC_jam]
Real-world use case for 3.10(a)
21:31:43 [DanC_jam]
21:32:37 [DanC_jam]
AndyS: most JDBC drivers deliver the whole result set before letting you get at rows...
21:32:58 [DanC_jam]
... because in, e.g. a 3-tier thingy, the query engine doesn't want to manage resources over time
21:33:12 [Yoshio]
q+ to ask why "sort and limit" seems to have no support here, it's so common in daily-life!
21:33:45 [AndyS]
AndyS: example of implications for JDBC (e.g. default in MySQL)
21:34:57 [DanC_jam]
ack robs
21:34:57 [Zakim]
RobS, you wanted to discuss implementation experience with no limit support
21:35:36 [AndyS]
Rob: Network Inference system does not have this feature
21:35:53 [DaveB]
what feature?
21:35:57 [AndyS]
Users terminate the query unsubtly
21:36:01 [AndyS]
21:36:11 [DanC_jam]
ack yoshio
21:36:11 [Zakim]
Yoshio, you wanted to ask why "sort and limit" seems to have no support here, it's so common in daily-life!
21:36:39 [HiroyukiS]
HiroyukiS has joined #DAWG
21:37:12 [AndyS]
Support for SORT: Simon worried about the cost of a sort
21:39:34 [AndyS]
Discussion on "sort" : tradeoff of client and server compute power per user
21:42:04 [AndyS]
Kendall: experience is that their apps UI wants "a few" results. It currently gets whole results before going on.
21:42:31 [Yoshio]
re: resource, what about "in-house" case?
21:42:51 [ericP]
q+ to invoke SQL and HTTP experience
21:45:42 [AndyS]
NetInf system is 3+-tier
21:45:57 [AndyS]
so added functionality in business logic
21:46:04 [AndyS]
21:46:22 [kendall]
for *web* apps, between arbitrary web clients and web servers, it seems that this 3 tier stuff is otiose
21:47:06 [AndyS]
Rob: Its a question of priorities
21:47:32 [AndyS]
Simon: LIMIT is trivially easy
21:47:50 [AndyS]
and useful
21:49:00 [kendall]
*and* likely to get done anyway
21:49:04 [howardk]
otiose? let's eschew obfuscation please!
21:52:42 [kendall]
kendall has joined #dawg
21:52:56 [DanC_jam]
DanC_jam has joined #dawg
21:53:12 [AndyS]
AndyS has joined #dawg
21:53:26 [AndyS]
Strawpoll on "Limit"
21:53:26 [AndyS]
All: as objective
21:53:26 [AndyS]
7 as requirement
21:53:26 [AndyS]
If we move it to be an objective: can Kendall craft text to explain why it is NOT an requirement
21:53:28 [HiroyukiS]
HiroyukiS has joined #DAWG
21:53:35 [ericP]
danc: limiting has an obvious syntergy with sort
21:53:47 [ericP]
danc: sorting is not *in*
21:54:06 [AndyS]
Connection between limit and sort should be noted
21:54:17 [ericP]
... implementations are deployed without it.
21:59:44 [Yoshio]
Yoshio has joined #dawg
22:00:04 [DaveB]
(limit sounded liked a rejected requirement to me from the irc. We have never rejected one yet, IIRC)
22:00:13 [Yoshio]
* sorry, I lost my guts :)
22:00:25 [AndyS]
We are still discussing it - no vote yet
22:00:50 [AndyS]
(we had a temp net glitch)
22:00:55 [DaveB]
I'd go with limit only, no sorting, no offset.
22:01:04 [AndyS]
22:02:32 [AndyS]
Req:7 Obj:2 Not at all:0
22:02:41 [AndyS]
Dave - as requriment or objective
22:03:03 [DaveB]
22:03:24 [howardk]
howardk has joined #dawg
22:04:41 [Yoshio]
* oh, I'm loosing the guts, again, it's too COOL here
22:05:24 [AndyS]
Vote: Network Inference formally objects
22:05:55 [ericP]
RobS (Network Inference) objects: below his threshold for what should be on critical path.
22:06:09 [AndyS]
ACCEPT: For: all \ Against: 1 Abst: 1
22:06:25 [ericP]
... falls below some other things (disjunction) that have been pushed to DO.
22:09:30 [AndyS]
-- Move to 3.11 Iterative Query
22:09:40 [AndyS]
Is this cursors?
22:11:36 [AndyS]
Data points: it is the comment from Chris Wilper
22:11:51 [AndyS]
22:12:20 [DanC_jam]
2.3 cursors and proofs seems relevant
22:12:29 [kendall]
daveb: do you support 3.11 iterative query?
22:12:36 [DaveB]
22:12:40 [kendall]
22:13:07 [AndyS]
Dan proposes that that there is no support at the current time.
22:13:27 [DanC_jam]
Simon: streaming results seems to cover this
22:14:43 [AndyS]
. ACTION: SimonR draft a reply to Chris Wilper and send draft to WG mailing list
22:14:59 [AndyS]
ACTION: SimonR draft a reply to Chris Wilper and send draft to WG mailing list
22:15:03 [Yoshio]
* Can't we control the temparature here? It's too cold for me
22:15:19 [JosD]
was about
22:17:11 [AndyS]
Dan suggests that the requirements list is closed-for-now: takes more effort to intro a new req
22:18:40 [AndyS]
-- Discussion on disjunction
22:19:00 [AndyS]
Simon: find people from NZ or Oz
22:19:58 [AndyS]
Move discussion to tomorrow
22:20:03 [AndyS]
-- Objectives
22:32:42 [Yoshio]
* Thank you very much, Rob!
22:32:57 [DanC_jam]
elaborating on closed-for-now... what I mean is: I'm not aware of any pending requirements. Before I'll spend group time on a requirement, you'll have to drum up a certain amount of support.
22:34:45 [AndyS]
22:34:52 [ericP]
oh yeah
22:35:01 [ericP]
4.1 Human-friendly Syntax
22:35:19 [ericP]
danc: impossible to quantify or test
22:35:40 [ericP]
[discussion of "easily"]
22:36:21 [ericP]
robS: i think it's useless without "easily"
22:36:59 [ericP]
simon: i think "easily" could be taken as an accesibility issue.
22:38:49 [ericP]
danc: trying to figure out if this is worht the screen real estate
22:39:13 [ericP]
... i think it's worth communicating this with our audience
22:40:02 [ericP]
RESOLVED: 4.1 accepted. no objection or abstentions
22:40:17 [ericP]
4.2 Provenance
22:41:06 [ericP]
implementation experience: 5 implementors
22:41:27 [ericP]
howard: natural piece of data i keep
22:42:53 [ericP]
ericP: without it, trust model would go away
22:43:42 [ericP]
kendal: our only paying client (intelligence community) wants it. lab would close up without it
22:43:45 [ericP]
22:44:30 [ericP]
kendalll: at query lang, you find a few kinds of designs. most common is quads.
22:45:20 [ericP]
Jos: if it's published on the web (has a URI), is that enough?
22:46:09 [ericP]
kendall: i believe so.
22:46:26 [ericP]
RobS: worried about defining provenance
22:46:43 [ericP]
... we don't think we've impelemented provenance
22:47:04 [ericP]
... RDF says [ reificiation process ]
22:49:42 [ericP]
AndyS: BRQL spec is not about provenance
22:49:49 [ericP]
... it's about data management.
22:50:20 [ericP]
... remove this model. replace with that model.
22:51:32 [ericP]
... next step is to offer that to the user (implies quads).
22:51:52 [ericP]
... next step is to offer it to the user in the QL
22:52:06 [ericP]
... every other impl of RDQL impelements quads.
22:52:13 [ericP]
... thinking about putting it into jena
22:52:39 [DanC_jam]
RobS's point about this being out of scope is fairly well made. The chair considers this provenance/source stuff to be on the edge of our charter.
22:52:43 [ericP]
kendall: we specifically need data management.
22:54:15 [ericP]
simon: we were a pure quads impl.
22:55:09 [ericP]
... for symmetry, we offered that to the QL.
22:55:28 [ericP]
... we don't use it terribly often
22:55:36 [ericP]
... we use it for security.
22:55:55 [ericP]
... it would be insane securing individual models.
22:56:01 [ericP]
22:58:36 [AndyS]
ACTION: Andy+Kendall, reword 4.2
22:58:52 [ericP]
andyS: not sure what to put in for inferred triples
22:58:54 [DaveB]
how reword?
22:59:08 [DaveB]
I don't want no fuzzy 4th part of a "triple"
22:59:10 [AndyS]
Remove provenace: more about data management and source
22:59:59 [ericP]
ACTION kendall, andy: write up a new wording of 4.2 Provenance that focuses on data management
23:00:43 [ericP]
strawman results: support except for robS
23:02:03 [DaveB]
23:02:35 [ericP]
folks would support 4.2 as a DO if it was rephrased per kendall and andy's action item
23:02:38 [AndyS]
We had a strawpoll to test the water
23:02:59 [ericP]
it was as warm as bath water
23:02:59 [DaveB]
okay. you'll also be reading the logs sometime and wondering what you recorded ;)
23:03:43 [AndyS]
Thanks for pointing out the unclarity
23:10:27 [kendall]
it would help jos to add "in the queried graph" to 4.3, which I undertake to do.
23:12:21 [AndyS]
This only makes sense only for variable bindings
23:17:25 [ericP]
danc: 4.3 Non-existent Triples will cost the working group
23:17:42 [ericP]
robS: we will implement it one way or another
23:18:59 [ericP]
RESOLVED: "4.3 Non-existent Triples" accepted as a DO over abstentions from AndyS and Jos
23:29:11 [DanC_jam]
ack ericp
23:29:11 [Zakim]
ericP, you wanted to invoke SQL and HTTP experience
23:30:57 [Yoshio]
q+ to say how one implement this is another thing
23:34:56 [Yoshio]
* and this is not a requirement but an objective design
23:37:00 [DanC_jam]
ack yoshio
23:37:00 [Zakim]
Yoshio, you wanted to say how one implement this is another thing
23:47:48 [ericP]
ACTION simon: write a union query version
23:48:44 [ericP]
4.5 Aggregate Query tabled for now
23:49:58 [ericP]
danc: how many people like "4.9 Boolean Results" as written
23:50:06 [ericP]
23:50:07 [ericP]
23:50:43 [ericP]
danc: do you not yes or know queries?
23:50:54 [ericP]
robS: yes
23:51:22 [DanC_jam]
hmm... how about: yes/no questions should be straightfoward to express.
23:51:23 [Yoshio]
know queries?
23:52:51 [ericP]
yoshio: we can't adopt this without adopting non-existent triples
23:55:45 [ericP]
simon: not happy with boolean nature
23:55:56 [Yoshio]
sorry, but I remember we ADOPTED 4.3!
23:56:41 [ericP]
Resolved: accept "yes/no questions should be straightfoward to express"
23:58:02 [ericP]
ACTION Kendall: remove 4.4 User-specifiable Serialization