Chatlog 2010-03-09

From SPARQL Working Group
Jump to: navigation, search

See original RRSAgent log and preview nicely formatted version.

Please justify/explain all edits to this page, in your "edit summary" text.

14:50:26 <RRSAgent> RRSAgent has joined #sparql
14:50:26 <RRSAgent> logging to
14:50:28 <trackbot> RRSAgent, make logs world
14:50:28 <Zakim> Zakim has joined #sparql
14:50:30 <trackbot> Zakim, this will be 77277
14:50:30 <Zakim> ok, trackbot; I see SW_(SPARQL)10:00AM scheduled to start in 10 minutes
14:50:31 <LeeF> zakim, this will be SPARQL
14:50:31 <trackbot> Meeting: SPARQL Working Group Teleconference
14:50:31 <trackbot> Date: 09 March 2010
14:50:32 <Zakim> ok, LeeF; I see SW_(SPARQL)10:00AM scheduled to start in 10 minutes
14:50:33 <LeeF> Chair: LeeF
14:50:38 <LeeF> Scribe: Souri
14:50:44 <LeeF> Agenda:
14:53:11 <AlexPassant> AlexPassant has joined #sparql
14:55:36 <Zakim> SW_(SPARQL)10:00AM has now started
14:55:42 <Zakim> +??P0
14:55:47 <AndyS> zakim, ??P0 is me
14:55:47 <Zakim> +AndyS; got it
14:57:08 <MattPerry> MattPerry has joined #sparql
14:58:17 <bglimm> bglimm has joined #sparql
14:58:26 <Zakim> +MattPerry
14:59:15 <Zakim> +kasei
14:59:23 <kasei> Zakim, mute me
14:59:23 <Zakim> kasei should now be muted
15:00:09 <ivan> zakim, dial ivan-voip
15:00:09 <Zakim> ok, ivan; the call is being made
15:00:11 <Zakim> +Ivan
15:00:30 <tommik> tommik has joined #sparql
15:00:34 <Zakim> +bglimm
15:00:42 <bglimm> Zakim, mute me
15:00:42 <Zakim> bglimm should now be muted
15:01:09 <Zakim> + +035840564aaaa
15:01:19 <tommik> zakim, aaaa is me
15:01:19 <Zakim> +tommik; got it
15:01:26 <Zakim> +LeeF
15:01:40 <LeeF> zakim, who's on the phone?
15:01:40 <Zakim> On the phone I see AndyS, MattPerry, kasei (muted), Ivan, bglimm (muted), tommik, LeeF
15:01:59 <AndyS> Hi all.
15:02:17 <Zakim> +??P19
15:02:21 <AlexPassant> Zakim, ??P19 is me
15:02:21 <Zakim> +AlexPassant; got it
15:02:34 <AxelPolleres> AxelPolleres has joined #sparql
15:02:38 <LeeF> Scribe: MattPerry
15:02:41 <LeeF> Scribenick: MattPerry
15:02:46 <LeeF> Regrets+ Souri
15:03:01 <Zakim> +dcharbon2
15:03:18 <LeeF> PROPOSED: Approve minutes at
15:03:59 <LeeF> RESOLVED: Approve minutes at
15:04:23 <SteveH> SteveH has joined #sparql
15:04:32 <SteveH> hi all
15:04:34 <LeeF> Next meeting: 2010-03-16 @ 14:00 UK / 10:00 EDT
15:05:04 <ivan> possible regrets for me
15:05:26 <MattPerry> LeeF: 2 weeks of 4 hr difference
15:05:31 <Zakim> + +0207735aabb
15:05:43 <SteveH> Zakim, aabb is me
15:05:43 <Zakim> +SteveH; got it
15:06:49 <MattPerry> LeeF: may be a problem that Paul is not here ... but we will try to make decisions anyway
15:07:13 <LeeF>
15:07:27 <MattPerry> Topic: Blank Nodes in Delete
15:08:01 <MattPerry> Option 1: no blank nodes in delete template
15:08:16 <MattPerry> Option 2: blank nodes as wildcard in delete template
15:09:56 <MattPerry> LeeF: if prohibit blank nodes now can we change it later
15:10:16 <Zakim> +AxelPolleres
15:10:47 <bglimm> Zakim, unmute me
15:10:47 <Zakim> bglimm should no longer be muted
15:11:09 <AndyS> I don't understand why if we exclude bnode syntax now we are effectively deciding for the future.
15:11:22 <LeeF> AndyS, yes, that's what I was trying to say
15:11:28 <MattPerry> bglimm: can live with no blank nodes but would be better with them
15:11:31 <ivan> q+
15:11:37 <LeeF> ack ivan
15:12:26 <SteveH> q+
15:12:31 <bglimm> q+ to ask about deleting lists without allowing bnodes
15:12:33 <LeeF> ack SteveH
15:12:34 <MattPerry> ivan: there are non-entailment related use cases that need blank nodes
15:13:30 <LeeF> DELETE { ?x :hasList (1 2 3) } WHERE { ... ?x }
15:13:35 <AxelPolleres> (just to be able to follow... what is discussed - said by ivan - is that not allowing to "delete bnodes" would  not allow to delete lists?
15:14:24 <LeeF> ack bglimm
15:14:24 <Zakim> bglimm, you wanted to ask about deleting lists without allowing bnodes
15:14:45 <MattPerry> SteveH: rdf list blank node shortcut doesn't add new functionality just makes it easier
15:14:48 <ivan> q+
15:14:50 <bglimm> ack bglimm
15:14:58 <AxelPolleres> what does this mean if the data graph has two lists (1 2 3) as value of :hasList ? (with different bnode ids?)
15:15:41 <AxelPolleres> q+
15:16:08 <ivan> ack ivan
15:16:15 <AxelPolleres> q+ to ask whether the vars in the DELETE template don't need to be bou�nd in the WHERE 
15:16:36 <LeeF> ack ivan
15:16:41 <SteveH> q+
15:17:16 <AxelPolleres> ack axel
15:17:16 <Zakim> AxelPolleres, you wanted to ask whether the vars in the DELETE template don't need to be bou�nd in the WHERE
15:17:17 <LeeF> ack AxelPolleres
15:17:50 <LeeF> ack SteveH
15:17:59 <MattPerry> LeeF: long var-based list expression must be repeated in both Delete and Where
15:18:40 <MattPerry> SteveH: can you use property paths to get list variables?
15:18:48 <AxelPolleres> so, just to note, such a delete would indeed all the matching lists (answering my own question from further above)
15:19:37 <LeeF> straw poll: preference between (1) prohibiting blank nodes in DELETE templates and (2) blank nodes in DELETE templates act as wild cards
15:19:44 <ivan> q+
15:19:54 <AndyS> Wheer are we on need "all triples must match" rule -- else chaos may result (half a list goes if wrong length)
15:19:55 <LeeF> ack ivan
15:20:12 <MattPerry> ivan: also same blank node cannot be in both delete and where
15:20:15 <kasei> q+
15:20:18 <AndyS>  Pref: 2
15:20:20 <bglimm> 2
15:20:23 <kasei> Zakim, unmute me
15:20:23 <Zakim> kasei should no longer be muted
15:20:23 <ivan>  Pref: 2
15:20:25 <LeeF> ack kasei
15:20:32 <SteveH>  pref: 1, but 2 probably ok
15:20:37 <MattPerry>  Pref: 2
15:21:13 <ivan> q+
15:21:17 <MattPerry> kasei: can rdf list be a special case?
15:21:23 <AxelPolleres> slight pref 2, still don't feel sure about implications
15:21:55 <MattPerry> kasei: grammar can enforce this
15:22:11 <LeeF> ack ivan
15:22:13 <MattPerry> LeeF: specification still needs to give semantics for blank nodes
15:22:25 <kasei> Zakim, mute me
15:22:25 <Zakim> kasei should now be muted
15:22:38 <LeeF> My preference is for #1
15:22:50 <MattPerry> ivan: many cases for blank nodes beyond rdf lists
15:23:00 <AxelPolleres> q+ to note that conceptually, it seems strange, since it would be the only place where bnodes have a "universal variables" meaning, whereas anywhere alse they rather indicate "existential variables"
15:23:19 <AxelPolleres> ... but I have no better proposal
15:23:27 <LeeF> ack AxelPolleres
15:23:27 <Zakim> AxelPolleres, you wanted to note that conceptually, it seems strange, since it would be the only place where bnodes have a "universal variables" meaning, whereas anywhere alse they
15:23:27 <AndyS> but they are bnodes in the data.  a ?var can be bound to a data bnode
15:23:31 <Zakim> ... rather indicate "existential variables"
15:24:00 <bglimm> I think in my examples you could always use the [] notation and as I understand it, that would be allowd even under 1
15:24:42 <bglimm> and Steve
15:25:06 <AndyS> opt 1 bans [] in delete template AIUI
15:25:07 <SteveH> I'm about -0.5
15:25:30 <LeeF> PROPOSED: Blank nodes in DELETE templates act as "wild cards", effectively as variables bound to all RDF terms; the same blank node cannot be used in the WHERE clause and the template, or in multiple BGPs
15:26:03 <kasei> +1
15:26:05 <ivan> second it
15:26:06 <bglimm> +1
15:26:07 <ivan> +1
15:26:16 <SteveH> abstain
15:26:20 <dcharbon2> abstain
15:26:42 <LeeF> RESOLVED: Blank nodes in DELETE templates act as "wild cards", effectively as variables bound to all RDF terms; the same blank node cannot be used in the WHERE clause and the template, or in multiple BGPs, SteveH, dcharbon2, LeeF abstaining
15:26:50 <AxelPolleres> +1 (lacking better solutions)
15:27:28 <LeeF>
15:27:28 <MattPerry> Topic: Data Sets in SPARQL Update
15:27:40 <SteveH> [[[
15:27:41 <SteveH> something like:
15:27:43 <SteveH> DELETE { ?n rdf:first ?f ; rdf:rest ?n }
15:27:43 <SteveH> WHERE { <list> rdf:first*/rdf:rest* ?n . ?n rdf:first ?f ; rdf:next ?n }
15:27:44 <SteveH> ]]]
15:30:07 <MattPerry> LeeF: SPARQL query allows specification of an RDF Dataset, but SPARQL update does not allow query to select specific RDF Dataset
15:30:33 <SteveH> q+
15:30:49 <AndyS> Is this different from using GRAPH in the pattern?
15:31:00 <SteveH> GRAPH ?g ... FILTER(?g = <a> || ...)
15:31:24 <MattPerry> LeeF: what is the scope of graphs that the WHERE is matched against?
15:31:35 <AndyS> If the default dataset is changed (to some union using FROM...), it goes make a difference.
15:32:19 <MattPerry> LeeF: No way to give default graph for update
15:32:28 <AndyS> q+ to ask how does this interact with WITH?
15:32:30 <LeeF> q?
15:32:33 <LeeF> ack SteveH
15:33:02 <ivan> Lees' syntax example: INSERT INTO <g1> { template } FROM g2 FROM g3 FROM NAMED g4 FROM NAMED 
15:33:02 <ivan> g5 WHERE { GRAPH ?g { ?s ?p ?o } }
15:33:10 <LeeF> SELECT ... FROM <g1> FROM <g2> { tp1 . tp2 }
15:33:51 <LeeF> SELECT ... { GRAPH ?g1 { tp1 } . GRAPH ?g2 { tp2 } }
15:34:16 <MattPerry> LeeF: can do this with GRAPH but WHERE gets very complicated
15:34:17 <AndyS> not quite - tp may span g1 and g2.
15:34:34 <LeeF> AndyS, one tp can span two graphs?
15:34:37 <LeeF> q?
15:34:39 <LeeF> ack AndyS
15:34:39 <Zakim> AndyS, you wanted to ask how does this interact with WITH?
15:34:57 <AxelPolleres> SELECT ... { GRAPH ?g1 { tp1 } . GRAPH ?g2 { tp2 } } FILTER (?g1 = <g1> or ?g1= <g2> and ?g2 = <g1> or ?g2= <g2> )
15:34:59 <AxelPolleres> ?
15:35:36 <LeeF> AxelPolleres, right
15:35:37 <AndyS> tp1 can match g1 or g2, tp2 can match g1 or g2 => 4 cases
15:35:37 <LeeF> q?
15:35:50 <AndyS> (bnodes ... :-))
15:36:12 <AxelPolleres> q+
15:36:22 <AndyS> I'm happy to consider the design.  Seems harmless (so far).
15:36:54 <AndyS> I can see that WITH != FROM (WITH is the updated graph, FROM is the queried graph)
15:37:08 <LeeF> ack AxelPolleres
15:38:12 <AndyS> Can't update a synthetic graph (e.g. merge of 2 graphs)
15:38:27 <LeeF> q?
15:38:39 <AndyS> LeeF, your suggestion is good.
15:39:01 <MattPerry> LeeF: interaction of WITH and FROM needs more investigation
15:39:16 <AxelPolleres> I see a problem with e.g. INSERT { ?X p ?Y } FROM <g1> FROM <g2> WHERE { ?X p1 o1 . ?Y p2 o2. }
15:40:31 <AxelPolleres> ... don't know what it means. A proposal should make this corner cases clear, before we can really figure out whether/where it is useful.
15:41:09 <LeeF> ACTION: Lee to work with Paul to flesh out design proposal for FROM/FROM NAMED (datasets) in SPARQL Update
15:41:09 <trackbot> Created ACTION-202 - Work with Paul to flesh out design proposal for FROM/FROM NAMED (datasets) in SPARQL Update [on Lee Feigenbaum - due 2010-03-16].
15:41:45 <LeeF>
15:42:00 <MattPerry> Topic: Update Fault Types
15:42:55 <MattPerry> AndyS: with HTTP there are few error codes
15:44:03 <MattPerry> dcharbon2: WSDL 2.0 has no limit on error codes
15:44:06 <dcharbon2>
15:45:41 <Zakim> -AlexPassant
15:45:43 <SteveH> q+
15:46:09 <LeeF> ACK SteveH
15:49:07 <Zakim> +??P2
15:49:08 <AlexPassant> Zakim, ??P2 is me
15:49:08 <Zakim> +AlexPassant; got it
15:49:22 <MattPerry> LeeF: does it make sense to have a lot of faults if there are only a couple of error codes?
15:50:29 <MattPerry> dcharbon2: could start with 2 basic error codes and see what users think
15:50:30 <kasei> i'd be interested in seeing the use of many of the 2xx http codes be conformant (going beyond the idea of just faults). 202 in particular.
15:51:23 <MattPerry> LeeF: what code does drop non-existent graph map to?
15:51:26 <kasei> 400 bad request seems like a potential status code...
15:51:56 <AndyS> "202 Accepted" is not an error? 2xx are all positives? Text is "Successful 2xx"
15:52:06 <AndyS> ?? "304 Not Modified"
15:52:13 <SteveH> could be
15:52:20 <MattPerry> SteveH: there are many more-specific HTTP codes to use
15:53:10 <MattPerry> LeeF: lets hold off on error codes until update language is set
15:54:02 <MattPerry> AndyS: it is still useful to identify what the possible errors are even though they are not exposed in the SPARQL protocol
15:54:14 <SteveH> +1 to AndyS
15:54:21 <MattPerry> AndyS: informative text can do this
15:55:35 <MattPerry> LeeF: Next week HTTP protocol, Property Paths, what is and is not in update language, and F2F
15:55:49 <Zakim> -bglimm
15:55:50 <Zakim> -SteveH
15:55:50 <Zakim> -LeeF
15:55:50 <Zakim> -AxelPolleres
15:55:53 <Zakim> -kasei
15:55:54 <Zakim> -MattPerry
15:55:56 <Zakim> -dcharbon2
15:55:56 <ivan> zakim, drop me
15:55:56 <Zakim> Ivan is being disconnected
15:55:58 <Zakim> -Ivan
15:55:58 <Zakim> -tommik
15:56:12 <SteveH> it's spooky that zakim knows more of my phone umber than I do :)
15:56:20 <AndyS> zakim, drop me
15:56:20 <Zakim> AndyS is being disconnected
15:56:22 <Zakim> -AndyS
15:58:52 <kasei> can anyone help me parse a sentence from the RDF/XML spec?
15:59:13 <LeeF> kasei, can try :D
15:59:25 <kasei> i'm looking for a URI that identifies RDF/XML, the serialization format.
15:59:33 <kasei> section 5.1 says, "The RDF Vocabulary is identified by this namespace name"
15:59:42 <kasei> I'm trying to sort out if "RDF Vocabulary" is what I'm after.
15:59:51 <kasei> I suspect not, but not totally sure.
15:59:56 <AndyS> no, it's not.
16:00:10 <AndyS> RDF vocab is "rdf:type" etc.
16:00:23 <kasei> bah. annoyed that we've been able to get this far without URIs for some very basic stuff!
16:00:32 <AndyS> Not sure there is a URI for the synatx - is there a URI for every MIME type?
16:01:09 <AndyS> There is naming competition between MIME types and formats ... so not sure if anyone has been bold enough to go there.
16:01:10 <kasei> well, much like the rdf/xml spec, there's probably an offical information resource for mime types, but that's different.
16:01:27 <AndyS> Fairly certain W3C hasn't - would be a nice suprise if they had.
16:01:30 <AndyS> ivan?
16:02:13 <kasei> wonder if this is the sort of thing that might be added to rdf/xml based on the upcoming workshop...
16:02:43 <AndyS> kasei, Graham Klyne would be a good person to ask - he tracks IETF and W3C.
16:02:50 <kasei> if we go ahead with this saddle:resultFormat stuff in the SDs, I'd like to be able to point to URIs for the standard formats.
16:03:06 <SteveH> kasei, didn't saddle: use mime types?
16:03:31 <kasei> it used both mime types and a link to the spec's webpage.
16:03:37 <SteveH> right
16:03:47 <SteveH> given that conneg works on mime types, that's not a bad idea
16:03:53 <kasei> which is fine if you don't have a proper URI, I suppose, but I think we can/should do better.
16:05:19 <kasei> also, with the confusion around some mime types for rdf, I'd rather not lean too heavily on them as identifiers.
16:05:47 <SteveH> true
16:06:23 <Zakim> -AlexPassant
16:06:23 <kasei> especially n-triples. text/plain isn't exactly the most useful thing for conneg.
16:06:24 <Zakim> SW_(SPARQL)10:00AM has ended
16:06:25 <Zakim> Attendees were AndyS, MattPerry, kasei, Ivan, bglimm, +035840564aaaa, tommik, LeeF, AlexPassant, dcharbon2, +0207735aabb, SteveH, AxelPolleres
16:09:21 <AndyS> agree re text/plain.  They didn't plan to let it out of the WG as a format - but it escaped.  Feral format.
16:09:40 <kasei> heh
16:47:00 <SteveH> SteveH has joined #sparql
17:38:03 <AndyS> AndyS has joined #sparql