12:28:55 RRSAgent has joined #ldp 12:28:55 logging to http://www.w3.org/2013/09/13-ldp-irc 12:28:56 RRSAgent, make logs public 12:28:56 Zakim has joined #ldp 12:28:58 Zakim, this will be LDP 12:28:58 ok, trackbot; I see SW_LDP()8:30AM scheduled to start in 2 minutes 12:28:59 Meeting: Linked Data Platform (LDP) Working Group Teleconference 12:29:00 Date: 13 September 2013 12:31:08 rgarcia has joined #ldp 12:43:39 SW_LDP()8:30AM has now started 12:43:46 +Workshop_room 12:44:36 +SteveBattle 12:44:58 stevebattle4 has joined #ldp 12:45:08 davidwood has joined #ldp 12:45:35 nmihindu has joined #ldp 12:46:58 Good morning everybody. 12:47:37 scribenick: nmihindu 12:48:14 Arnaud1: we can spend the morning discussing the status about other working group documents 12:48:39 topic: Use Cases & Requirements 12:48:55 https://dvcs.w3.org/hg/ldpwg/raw-file/default/ldp-ucr.html 12:49:25 stevebattle4: I got valuable comments from mestaban and most of them are addressed now 12:49:31 http://lists.w3.org/Archives/Public/public-ldp-wg/2013Sep/0029.html SteveB's update 12:50:10 stevebattle4: use cases are numbered correctly now and they can be referenced 12:51:13 stevebattle4: the email I sent includes how I addressed miguel's comments 12:51:59 stevebattle4: the only comment that was not addressed was changing the user contributed user stories 12:53:15 ... : miguel was asking for informative references and I've added for some of the use cases 12:54:41 ... : one of the problems was to add local references 12:55:08 Arnaud: davidwood has a action item on reviewing the UCR 12:55:14 +??P9 12:55:28 davidwood: I've already gave my reviews 12:55:50 davidwood: do you want me to review a specific section ? 12:56:00 zakim, ??P9 is me 12:56:00 +rgarcia; got it 12:56:11 Only section 5 needs a robust review 12:56:57 Arnaud: how long does it take to finish the remaining work to publish an update ? 12:57:33 stevebattle4: I just have to finish the references, so I can finish it by next meeting 12:58:12 Arnaud: If we can finish it within a week or two, it would be great 12:58:29 topic: Best practices and guidelines document 12:58:41 Arnaud: what is the current status ? 12:59:05 cody: we have formalized all that we had in the previous wiki 12:59:41 ... I have polished in for grammar and everything 13:00:01 ... just have to finish adding the references 13:00:12 Arnaud: is it in respec format ? 13:00:15 cody: yes 13:00:39 https://dvcs.w3.org/hg/ldpwg/raw-file/default/ldp-bp/ldp-bp.html 13:01:00 Arnaud: can someone volunteer to review it ? 13:01:19 action: steve to review the Best Practices & Guidelines 13:01:19 'steve' is an ambiguous username. Please try a different identifier, such as family name or username (e.g., sbattle2, sspeiche). 13:01:28 action: steves to review the Best Practices & Guidelines 13:01:28 Created ACTION-95 - Review the best practices & guidelines [on Steve Speicher - due 2013-09-20]. 13:01:42 action: miguel to review the Best Practices & Guidelines 13:01:42 Created ACTION-96 - Review the best practices & guidelines [on Miguel Esteban Gutiérrez - due 2013-09-20]. 13:01:45 TallTed has joined #ldp 13:01:45 mesteban and SteveS volunteer to review it 13:02:36 Arnaud: Once we finish the review we can publish it 13:02:54 Ashok has joined #ldp 13:06:12 Arnaud: if anyone want to add any section to best practices guidelines document, they can contribute it and we can discuss it in the WG 13:06:54 topic: Test suite 13:07:21 Arnaud: what is the progress ? 13:07:53 ... raul has raised the issue of lack of MUSTs that are testable 13:08:44 ... we discussed yesterday that separating restricted / unrestricted LDPRs, so some SHOULDs will become MUST 13:08:57 ... that will have an impact on the test suite 13:09:07 ... we will have more testable requirements 13:09:21 rgarcia: I don't see any problem in it 13:09:37 SteveS: why the SHOULDs can not be tested ? 13:09:59 rgarcia: we can test them but we will have to find a way to do the grouping 13:10:17 ... because they only apply in some scenarios such as paging 13:11:14 JohnArwe: are you talking about something like modules ? you can do that in the test suite even though the spec does not have that modularity 13:11:58 rgarcia: that can be done but the it is better if that categorization is in the spec too 13:12:44 Ashok: shoulds are somewhat optional. How can we test that ? 13:13:14 ... what happens if a client/server doesn't do SHOULDs, does that mean it fails the test ? 13:13:45 davidwood: Must are pass/fail, SHOULDs are pass or not present 13:14:12 JohnArwe: we should use the notion of extra credits for SHOULDs 13:14:42 Arnaud: failure of a SHOULD does not mean not compliance 13:15:47 Arnaud: what is the conclusion ? 13:16:02 ... we have a lot of suggestions for rgarcia 13:16:56 rgarcia: passing / failing SHOULDs do not give much information 13:17:32 ... we can have conformance and strict conformance 13:17:59 ... where one only consider MUSTs and other considering MUSTs and SHOULDs 13:18:25 JohnArwe: It is bit similar to schema conformance 13:18:39 Arnaud: people are interested in having some tests for SHOULDs 13:19:13 SteveS: in the implementations, we have tests for all features whether they are MUSTs or SHOULDs 13:19:27 Arnaud: do you need any help ? 13:20:02 rgarcia: Once the spec is updated, I can update the tests 13:20:58 topic: Access control note 13:21:30 Arnaud: what is the progress ? 13:22:03 Ashok: we have got one review from mesteban and waiting for the review from TallTed 13:22:22 TallTed: there is not much to review there 13:22:36 Ashok: what is the purpose of the access control note ? 13:23:21 Arnaud: Initially we wanted to address the security and address control in the LDP spec 13:23:35 ... but people complained it will take too much time 13:24:00 ... so we decided to look in to use cases and requirements and capture those as a WG note 13:24:23 ... and identify the possible solutions to address to these requirements 13:24:32 Ashok: is it a deliverable or a nice to have ? 13:25:05 Arnaud: It is a deliverable 13:26:08 ... it is useful to gather the requirements so if we decide to go for a recommendation track we already have background 13:26:28 ... and provide users some best practices 13:27:00 Ashok: are there any requirements that are special to LDP ? 13:27:56 sandro: there might be some. Eg. Some users can see a only a set of triples etc. 13:29:14 +bblfish 13:29:40 Ashok: When talking about read write, we have to talk to about access control. but the question is who is responsible for that. LDP or underlying database etc ? 13:30:38 Arnaud: to do expect the review from TallTed ? 13:30:48 hi 13:30:59 Ashok: I would be happy to get it reviewed by TallTed too 13:31:59 Arnaud: Ashok does not have to do everything from the scratch, people should contribute that 13:32:15 sandro: is it worth considering this for a workshop ? 13:32:28 Ashok: that is a good idea 13:33:13 The technology for what? 13:33:30 what is the problem? 13:33:32 Arnaud: If had this in the charter, we would have to come up with use cases and requirements 13:33:39 ... why can't we do it now ? 13:33:47 The use case for Access Control? 13:33:54 bblfish, yes 13:34:13 ah ok. YEs, we have the tech, for ID, Autheentication and Access control 13:35:08 Arnaud: do we have a need to specify access control at LDP level ? 13:35:34 JohnArwe: we can use technologies that is used by the Web in general 13:36:17 q+ 13:36:35 TallTed: we can address this in a different way 13:36:56 ... looking at what happens when there is no access control in place 13:37:06 ... rather than coming up with use cases 13:37:17 q+ 13:38:13 ack bblfish 13:38:20 Ashok: are LDP and access control mechanisms are two completely separate worlds or do we have to group them ? 13:38:39 http://www.w3.org/2012/ldp/wiki/AccessControl 13:39:42 http://www.w3.org/wiki/WebAccessControl 13:39:51 bblfish: when I respond to OPTIONS do I have to consider the user who is making the request 13:40:07 ... and only provide the methods they have access to 13:40:25 ... or is it a separate thing ? 13:40:59 ... is headers provide information or do we link to access control document ? 13:41:25 TallTed: currently we don't have links access control rules 13:42:13 bblfish: does the Allow headers mean I have access to do that operation ? 13:42:50 TallTed: OPTIONS just mean querying the server for functionality 13:43:12 ... it doesn't mean that you have access for those operations 13:43:31 ... it can be done with a separate header or other mechanism 13:44:00 Arnaud: sandro proposed that this document should be treated as a call for a workshop 13:44:21 ... similar to what ericP did for the RDF validation workshop 13:44:34 cody has joined #ldp 13:44:43 http://www.w3.org/2012/12/rdf-val/SOTA 13:45:44 Arnaud: may be can produce a similar document with state of the art 13:45:59 q? 13:46:06 ack stevebattle 13:46:55 stevebattle: rather than iterating the existing technologies, we can try more to relate it to LDP spec 13:47:24 I think it is a good idea to have the Use cases for a workshop 13:47:41 There are a few people who would be very interested in an WebACL workshop. 13:47:58 q+ 13:48:09 What we can usefully say about access control relate to the granularity of access, can we control individual resources or containers of resources. This might simply be a section of the best-practices document. 13:48:15 Arnaud: what can we standardize in this area ? 13:48:36 ... the only possible thing I see is access control to rdf 13:49:27 bblfish: one thing we can do is use LDP for specifying access control 13:50:06 TallTed: it is one option. It could be useful but it should not be mandatory 13:50:20 roger has joined #ldp 13:50:54 bblfish: if we don't use LDP for access control, how can a client know and specify access control ? 13:51:49 Arnaud: Ashok, how should we progress ? 13:52:01 so this means a client cannot know if what he is publishing is private, if it can be overwrittent, if anyone can see it, who can edit it, etc... 13:52:06 AndyS has left #ldp 13:52:08 so really seriously you NEED it 13:52:52 Ashok: there were some ideas such as fine grained access control. But do we need to integrate this with LDP, I am doubtful about it. 13:53:07 q+ 13:53:41 i think that this depends on client-driven or server-driven. I think Henry refers to client-driven access control, and I agree, I think it would be great for LDP to manage that. For server-driven access control, I would expect the server to drive it then .... but, then it would be good for the client to introspect that. 13:55:40 Cody: we can specify one layer of the security metadata in LDP but the implementations can implement in their own server specific way 13:55:45 ack bblfish 13:55:48 q+ to miguel 13:56:59 Arnaud: because LDP spec does not specify security mechanisms people will face problems how to add security 13:57:27 ... we can provide some guidelines on what to look at with this document 13:58:24 bblfish: we can specify the needs for LDP implementations and then come up with a workshop to resolve those 13:59:54 Arnaud: our plan is to provide a platform for the future work on this area 14:01:06 ack nmihindu 14:01:06 nmihindu, you wanted to miguel 14:01:17 TallTed: access control note can contain the concerns, current technologies, and limitations 14:02:00 mesteban: we should not focus too much on the current technologies but what are the differences in LDP scenarios 14:02:25 ... eg. when we follow links crossing boundaries what are the main concerns 14:02:56 Ashok: that is an interesting suggestion. Crossing boundaries make security hard. 14:02:59 We have the distributed Authentication, and Authorization, with 3 implementations I think now :-) 14:03:41 davidwood: distributed authentication is somehow solved but distributed authorization is still very hard though. 14:03:49 I can work with Ashok 14:05:02 mesteban: I would like to volunteer too 14:05:22 Ashok: how do we make progress ? 14:05:44 mesteban: for the first iteration we can work offline via email 14:06:06 I have implmented this wtih a a few people, so I have a much better idea that I had when I first contributed to this document 14:06:13 ok 14:06:51 -SteveBattle 14:07:04 Arnaud: we will have a break now, then look at the actions and move to the specification issues 14:07:09 Policy Awre Web paper: http://www.csee.umbc.edu/csee/research/swpw/papers/kolovski.pdf 14:08:19 http://dig.csail.mit.edu/2009/Papers/ISWC/policy-aware-reuse/cc.pdf 14:17:04 davidwood has joined #ldp 14:17:33 Arnaud: looking at open actions 14:17:35 s/Awre/Aware/ 14:17:45 Topic: Issue 77 14:19:19 Topic: Issue 83 14:19:27 Issue 77 is indeed covered in the Best Practices guide; you can resolve my action. 14:19:29 ... has been re-assigned to Roger 14:21:38 +SteveBattle 14:22:25 on Issue 47, this is an old request and there has been many changes to UC&R since. 14:22:37 re yesterday's Issue 81 discussion -- http://lists.w3.org/Archives/Public/public-ldp-wg/2013Sep/0035.html 14:23:41 hot discussions topics for today: PAGING, PATCH, membershipX naming, OPTIONS 14:23:50 Topic: OPTIONS 14:24:29 https://www.w3.org/2006/02/lc-comments-tracker/55082/ldp/2835?cid=2835 14:28:01 Is REST client or server driven? 14:30:23 SteveS has joined #ldp 14:31:18 I think of REST as a server-driven thing ... in my opinion. It is possible for the server to say to the client that "You can drive it in any way you think" - in which case it is 'server-driven client-driven' 14:34:20 OPTIONS is a cheaper version of GET is already agreed. 14:34:28 interesting relevant comment -- http://zacstewart.com/2012/04/14/http-options-method.html 14:39:30 discussion around other ways of achiving the same result, and if indeed these alternatives offer any real benifits 14:39:38 davidwood has joined #ldp 14:39:45 q+ 14:40:35 ack bblfish 14:41:25 Henry: yesterday was constraints on content, today is contraints on methods 14:42:45 Arwe: the constraints document could cover lots of differents types of constraining 14:43:55 TailTed: OPTIONs response has an single allow header which has a List of allowed methods 14:45:49 Arnaud: OPTIONs isn't is bad as people might think 14:46:13 yes, I think that is right: timbl seems to have the thought that one needed to do an OPTIONS before every GET . If that is not true then he'll probably be happy 14:46:59 Action: draft a response to Tim based on the consensus reached. 14:46:59 Error finding 'draft'. You can review and register nicknames at . 14:48:13 roger, who is that ACTION on? 14:48:40 q+ 14:48:40 it was a mistake to try and set it up as an official action ... Arnaud has assigned it in the comments tracker. 14:49:06 Topic: PAGING 14:50:50 John warns against tempting clients to depend upon information that they should not depend on 14:51:56 s/PAGING/Mark Baker's comment/ 14:54:54 Arwe: Mark Baker was objecting to clients being able to find out what profile a server supports, not objecting to LDP defining profiles or conformance classes 14:55:08 opic: Comment LC-2836: Paging 14:55:15 topic: Comment LC-2836: Paging 14:55:19 https://www.w3.org/2006/02/lc-comments-tracker/55082/ldp/2836 14:55:24 ack Ashok 14:55:41 Ashok: looking at how OData handles paging. 14:57:01 ... they use a Prefer header, so the client can add preferences. one of the preferences is max page size. this is interesting way of influencing the server 14:57:23 in rdf it's triples 14:57:30 +1 14:58:53 paging currently works by putting the "next" links in the document. Tim's suggestion was to put it into the header instead. 14:59:29 I thought we had links in the Headers 15:00:18 Q+ 15:00:30 bblfish, we do but don't recommend rel=next now 15:00:34 ack bblfish 15:01:48 bblfish: the client cannot influence what goes into the header, but, in theory they mght can able to do with wrt the content 15:04:12 perhaps you can do OPTIONS over pages, say you want the 5th page, without getting all the content 15:04:21 q+ 15:04:56 yes, agree. 15:05:05 you can move it down to http layer. 15:05:09 header are better for when the content is non-RDF 15:05:40 ack steves 15:05:41 ... and when the handling of this can actually be moved into HTTP client libraries 15:06:07 SteveS: atompub defines the paging headers 15:06:57 ... disadvantage of header-only, each page has its own URL too. 15:07:01 q? 15:09:05 q+ 15:09:18 difference between the full document, and then the pages which taken together are the full document 15:09:32 q+ 15:09:34 ack davidwood 15:11:32 davidwood: no violation of HTTP if we respond with pages, it's quite normal 15:12:25 ack SteveS 15:12:29 davidwood: sees arguments for links both in header and body 15:12:57 +q 15:13:15 q+ 15:13:58 q+ 15:14:11 ack roger 15:15:49 q+ to propose a mapping from the header to a *different* graph than the data payload 15:17:47 ack ashok 15:18:45 SteveS: it could be efficient in other ways to just look at headers to fetch next page without having needs to fire up an RDF parser and find next page 15:20:15 ack bblfish 15:20:32 TailTed: taking it out of the content is problematic, and putting it in both is fine. 15:20:36 q+ to propose some rationale for putting next page pointers only in the headers 15:22:13 bblfish: may important for telephones to inform the server what are their restrictions 15:22:27 s/may/maybe/ 15:23:00 ack ericP 15:23:00 ericP, you wanted to propose a mapping from the header to a *different* graph than the data payload 15:23:02 GET http://a.example/foo 15:23:02 218½ Location:http://a.example/foo1 Link:rel=next ref=http://a.example/foo2 15:23:03 => { [] a ldp:Page ; ldp:pageOf ; ldp:nextPage } 15:24:48 Content-Link: for the main page? 15:25:17 Is that second header not what Content-Link is for? 15:25:37 ack davidwood 15:25:37 davidwood, you wanted to propose some rationale for putting next page pointers only in the headers 15:25:43 sorry I meant Content-Location http://tools.ietf.org/html/draft-ietf-httpbis-p2-semantics-23#section-3.1.4.2 15:26:26 davidwood, doesn't want to polute the body with navigation links, would prefer to use the headers 15:28:47 q+ to talk about tabulator navigation 15:30:30 Opera Reader support for rel=next http://people.opera.com/howcome/2011/reader/ 15:31:08 ack ericP 15:31:08 ericP, you wanted to talk about tabulator navigation 15:31:09 Arnaud: in HTML, Link headers in to top of the HTML didn't really take off. 15:32:27 TailTed: have it in both places ! 15:32:46 I suppose the question of a MUST 15:33:25 if the MUST is in the header, then this does not exclude a possiblity of headers from the content ( thought then you have messy content ) 15:34:17 Ah yes, the edit issue on link relations is a good way of putting things. If you cannot edit it, because it is server logic, then it should be in the header. 15:35:48 erik: in a browser such as tabulator, don't want the control data and user data getting inter-mingled 15:36:03 s/erik/eric/ 15:37:29 sorry for mispelling you name EricP 15:37:43 s/mispelling/misspelling/ 15:38:56 Arnaud: mandatory headers, content optional 15:39:10 +1 15:39:26 +1 though I'd like to try to implement it and let you know 15:40:08 q+ 15:41:39 ack SteveS 15:41:51 Arnaud: we don't seem to have enough consistency about the control stuff which goes into content 15:44:28 strawpoll time !!!!! 15:44:58 STRAWPOLL: move page related links to HTTP headers 15:45:16 +1 15:45:39 +1 for mesteban 15:45:45 +0 I see both sides of the fence, different shades of green 15:45:51 +1 15:47:20 +0.7 ( I'll be fully +1 when I have implemented it ) And I think we need to allow it in the body for all the legacy clients that don't know about this header that don't know about this. 15:47:54 0 15:48:16 0 (not clear idea of the consequences) 15:48:24 +0.5 15:48:39 TallTed: 206 in HTTPbis might be the way to go 15:49:44 davidwood has joined #ldp 15:49:54 +0 15:49:55 +1 15:49:56 +1 15:50:10 0 15:50:15 deiu has joined #ldp 15:50:43 http://tools.ietf.org/html/draft-ietf-httpbis-p5-range-23 15:51:36 decision on this is postponed for now. 15:54:16 +0.5 15:54:37 john has agreed to look further into it 15:55:04 LUNCH BREAK. Resuming at 12:30 15:55:21 -rgarcia 16:04:04 http://tools.ietf.org/html/rfc5005#section-3 then page down to example 16:05:50 PhilA has joined #ldp 16:16:51 SteveS has joined #ldp 16:23:25 -bblfish 16:23:36 bblfish has joined #ldp 16:23:46 -SteveBattle 16:33:34 zakim, who's on the phone? 16:33:34 On the phone I see Workshop_room 16:38:22 topic: F2F logistics 16:38:49 Arnaud: possibility that we'll squeak through another LC review without needing another F2F 16:39:19 ... so it's wise to plan a F2F and decide later whether to cancel 16:39:28 ... david suggested an interop event 16:40:32 ... this would make the F2F more useful [if we don't have issues from LC comments] 16:41:44 ... sandro proposed meeting at TPAC in China 16:41:50 ... but may be too soon 16:42:15 AndyS has joined #ldp 16:42:25 AndyS has left #ldp 16:42:27 ... LC review period at least 3 weeks, after we've finished our issues, updated the draft, reviewed and published 16:52:06 davidwood has joined #ldp 16:52:33 JohnArwe has joined #ldp 16:53:28 PROPOSED: next (speculative) F2F 14-16 Jan 16:55:02 London would be available if needed 16:55:12 +0 16:55:24 +1 16:55:31 +0 16:56:42 Raleigh might be available 16:58:11 Nice/Sophia-Antipolis? 17:02:34 nmihindu has joined #ldp 17:06:32 http://tools.ietf.org/html/rfc5988#section-5.5 17:08:33 I don't think it's relevant to point to ATOM as an example because it expresses a specific schema. It is itself a defined resource type (a feed format), which in and of itself can have its own things like paging. But, open resources, in the case of LDP - that's a different story, isn't it? 17:09:46 cody, if i understand, i was making that point when i spoke about how Atom's link elements are within its controlled content model 17:10:22 am i correctly parsing your point? 17:10:33 OK. Xactly. I agree. Yes - I think we're talking about the same thing. 17:11:56 deiu has joined #ldp 17:14:27 +q 17:14:55 q+ 17:17:50 ack roger 17:18:54 q+ to discuss scoping of container triples vs. user data 17:19:11 roger: doesn't a container have user data? 17:19:51 ... looking at a bug tracker, it's not just an index; it can be anything 17:20:06 The LDP specification defines a Container as "a Linked Data Platform Resource (LDPR) representing a collection of same-subject, same-predicate triples." This can easily be misconstrued to mean that a Container should only contain same-subject, same-predicate triples. While Containers may contain only same-subject, same-predicate triples (i.e. the membership subjects and membership predicates of its membership triples), it is free to contain others. The definition is 17:20:07 meant to clarify only those attributes that are directly relavant to the interaction model of a Container, but not to limit them to those attributes alone. 17:20:07 It is important to remember that a Linked Data Platform Container (LDPC) is also a Linked Data Platform Resource (LDPR) and though it might exist as a membership controller, it may also represent additional data that is valuable to the agents that access it. 17:20:14 ... originally a container was just a link from container to resources 17:20:23 (That's from the BP, which I think clarifies that point about Containers) 17:20:35 ... i think the biggest polluter is all the membership-X stuff 17:21:26 ack steves 17:22:09 SteveS: sorting info is pretty different from the paging info 17:22:30 JohnArwe: sort criteria is in Section 5, not in paging 17:22:51 SteveS: when you get the data back, it's already sorted, so it would be odd to look in a link header 17:23:08 ... so it makes more sense to have sorting info in the container 17:23:18 ... so we just get rid of nextPage and pageOf 17:26:37 q- 17:31:29 action: ericP to send an email to the list describing what it would take to move page related triples to HTTP 17:31:29 Created ACTION-97 - Send an email to the list describing what it would take to move page related triples to http [on Eric Prud'hommeaux - due 2013-09-20]. 17:34:26 topic: Patch 17:34:34 Arnaud: we all wished we could have patch 17:34:44 ... we ended up saying it's optional 17:34:54 ... timbl objected to the optionality 17:35:17 ... he said "can't we lock people in a room for a weekend?" 17:35:23 ... we can: 17:35:45 ... .. 1 say welcome to our club of frustration 17:36:10 ... .. 2 make it mandatory without specifying the format X! 17:36:21 ... .. 3 quickly resolve a patch format 17:41:02 bblfish has joined #ldp 17:41:39 ericP: INSERT {bgp} DELETE {bgp} WHERE {bgp}, damnit 17:41:55 PROPOSED: INSERT {bgp} DELETE {bgp} WHERE {bgp}, damnit 17:43:04 JohnArwe has joined #ldp 17:43:49 Arnaud: do we need a WHERE? 17:44:11 ericP: i think we don't want to eliminate use cases with a BNode 17:44:26 davidwood: there may be other cases where you need a variable binding 17:45:12 +bblfish 17:45:14 ... we did this in calimacchus 17:45:15 -Workshop_room 17:45:16 +Workshop_room 17:45:44 Here was a simple approach I have for my bnode needs http://open-services.net/wiki/core/OSLC-Core-Partial-Update/#Example-update-blank-nodes-link-label 17:49:00 q+ 17:50:57 ack steves 17:51:33 JohnArwe: we have no SPARQL impls. our LDP impls are mapped to [conventional] SQL backends 17:52:03 ... also, PATH errata says not to re-use generic media types for patch 17:54:06 ericP: bgp with non-variable predicates is directly-translatable to SQL 17:54:23