W3C

- DRAFT -

RDB2RDF

03 Nov 2011

See also: IRC log

Attendees

Present
David, Seema, Michael, Richard, Ashok, Eric
Regrets
Chair
Ashok
Scribe
Ashok

Contents


ok

ok

Sure, Michael

Discussuing David's comments http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Oct/0022.html

All comments on R2RML document

* "The base IRI MUST be a valid IRI. It SHOULD end in a slash (“/”) character." - I would have thought this would say that it should end in an IRI delimiter, rather than specifically referring to a slash.

<dmcneil> URI spec: http://www.ietf.org/rfc/rfc3986.txt

David: Why not allow other valid IRI delimiters such as # -- ref to RFC3987

<dmcneil> IRI spec: http://www.ietf.org/rfc/rfc3987.txt

David: also change SHOULD to MUST

<dmcneil> (thanks seema)

<Seema> I agree with David's suggestions

David: Note on relative IRI resolution is unclear ... editors please improve wording

<Seema> http://www.w3.org/2001/sw/rdb2rdf/r2rml/

<mhausenblas> Michael: I've added Ashok's collected LC comments to http://www.w3.org/2001/sw/rdb2rdf/wiki/Last_Call

<mhausenblas> Michael: please use this space for notes and decisions (if we have a resolution, we can directly link it from there - same for new issues)

Thanks, Michael

<mhausenblas> Michael: Thanks to Ashok who did the hard work to collect them, I just pasted it there ;)

David: An R2RML data validator is a system that takes as its input an R2RML mapping, a base IRI, and a SQL connection to an input database, and checks for the presence of data errors. When checking the input database, a data validator MUST report any data errors that are raised in the process of generating the output dataset.

Ashok: Why do need this paragraph?

Seema and David: This is implementation guidance. Why do we need in the spec.

David: Section 41. * "SHOULD NOT contain any mapping components (logical tables, term maps, predicate-object maps) that are not referenced by a triples map (in other words, are “unused”)" - I think the list of mapping components should either be complete, or formally defined elsewhere. If I am reading it correctly, it omits join condition and referencing object map. This seems like a needless lack of clarity in the spec.
... define mapping components formally and completely and then refer in various palces in the spec.

Seema: +1

david: * "MAY assign IRIs or blank node identifiers to any mapping component in order to enable re-use of mapping components within the mapping graph." - I see that the term "mapping component" is used again here, so maybe it does warrant a formal definition. I see the term "mapping constructs" is used just below this. Are these the same thing?
... change mapping construct to mapping component

* "For example, an IRI that represents a subject map may be used as the subject map of multiple triples maps; and may even be used as an object map of another triples map if it has the right properties." - I notice that this is not an exhaustive list of how resources may be reused. Since it says "For example", does this count as non-normative? I only raise these questions. I am not enough of a spec-lawyer to know how best to handle this.

David: Show the example using the example syntax
... * "Using these classes is OPTIONAL in a mapping graph." - I think this wording could lead to confusion as it could be read to mean that an R2RML mapping could be defined without a TriplesMap. I think what is meant, as the next sentence clarifies, is that the triples defining the types of the mapping components are optional.
... what it is really saying is that you don't have to have rdf:type predicate on your resources. This needs to be clarified

<ericP> I'll be able to join in 1 hr

Seema: Mapping may not define all classes ... needs to be clarified

David and Seema: Wording needs to be clarified

David: * Since it doesn't say that an R2RML mapping graph MUST contain at least one TriplesMap, the implication is that an empty mapping graph that produces no triples is valid R2RML. I just mention this to confirm that is the intent.
... Clarify that a NULL mapping (without a triples map) is still a valid mapping
... * "The applicable class of a resource can always be inferred from its properties." - I think this captures a requirement on implementors that should be made more explicit. To get started processing an R2RML mapping, implementors must infer that a resource is a TriplesMap based on the presence of any of the mapping predicates which are required and have a domain of TriplesMap. Once you have the TriplesMaps then the rest of the types are implied by thei
... Call out type inferencing required for triples maps. The others are easier becuase they are also doing type inferencing but the position clarifies this.
... Section 4.2 * "The preferred file extension is .ttl." - Is preferred a weaker form of SHOULD?

Shall we change "preferred to "SHOULD"

Seema: Can Turtle files have other file extensions?

David: * "It is common to use document-local IRIs in mapping documents by defining the default prefix in the beginning of the document, and using it for creating IRIs for mapping components such as triples maps" - This note does not seem like it belongs in the spec.

David/Seema/Ashok: We recommend removing the paragraph

David: Section 4.3 * "A data error is a condition of the data in the input database that would lead to the generation of an invalid RDF term, such as an invalid IRI or an ill-typed literal." - This seems like quite a loose definition. Since we have a more formal definition just below, can we strike this?
... remove phrase "such as ..."

Seema: Agree ... remove phrase

David: * "A term map with a datatype override produces an ill-typed literal of a supported RDF datatype." Where "ill-typed" is defined in section 10 as "A typed literal of a supported RDF datatype is ill-typed if its lexical form is not in the lexical space of the RDF datatype identified by its datatype IRI." - Doesn't this definition of a mandatory error contradict our stated intent of: if data values fall outside of the intersection of SQL data types and X

Ashok: Suggest we defer this till we resolve issue 69

David: * "A logical table is a possibly virtual database table that is to be mapped to RDF triples. A logical table is either..." - I would remove the loose definition of "a possibly virtual database table" and combine these two sentences into one. The first time I read the sentence I thought it was talking about database views and database tables.
... Why a loose definition followed by an exact definition ... suggest we collapse into one
... Section 5.1 * "A SQL base tables or views is represented by a resource that has exactly one rr:tableName property." - change "tables"->"table" and "views"->"view"
... this is a typo
... * "If no catalog or schema are specified" - change "are"->"is"
... fix grammar
... * "If no catalog or schema are specified, then the default catalog and default schema of the SQL connection are assumed." - I think this is potentially confusing because it could be read to suggest that the R2RML implementor is supposed to do something wtih the default catalog/schema. But in fact the intent is for the identifier to be passed unchanged across the SQL connection.
... The R2RML does not have to do anything ... the default schema/catalog are sent automatically out of band
... make clear the R2RML processor does not have to do anything just send the unqualified identifier on through
... Section 5.2 * "A SQL query is a SELECT query in the SQL language that can be executed over the input database." - How about explicitly saying that it is a string representation of a SELECT query?
... wording could be improved

Seema: Maybe he means the concept not a string

<dmcneil> so maybe change the sentence defining the rr:sqlQuery object to be "a string that conforms to the production..."

Break for 10 minutes

<mhausenblas> http://www.w3.org/2001/sw/rdb2rdf/wiki/Last_Call

<cygri> david's comments: http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Oct/0022.html

<Seema> \me will be back in 10 min

<mhausenblas> Michael: Thanks David for adding your latest LC comment (16) to http://www.w3.org/2001/sw/rdb2rdf/wiki/Last_Call

David: * "It MUST NOT have duplicate column names or unnamed derived columns in the SELECT list." - It seems obvious now but when I first read this I had trouble understanding what was meant by "unnamed derived columns". I think it means if the query includes expressions as projected columns then they must be named.
... can we improve the wording
... * "For any database objects referenced without an explicit catalog name or schema name, the default catalog and default schema of the SQL connection are used." - Same comment as from previous section. I think this statement could be improved to clarify that an R2RML processor does not need to do anything to accomplish using the default catalog/schema.
... please clarify that happens automatically by connection and database
... Section 6 * "specifies how to generate the subjects for each row of the logical table" - I think this would be clearer if it said "generate a subject for each row". Otherwise it sounds like it might be possible for a subject map to generate multiple subjects from each row.

<cygri> David: * "specifies how to generate the subjects for each row of the logical

<cygri> table" - I think this would be clearer if it said "generate a subject for

<cygri> each row". Otherwise it sounds like it might be possible for a subject map

<cygri> to generate multiple subjects from each row.

Richard: Yes, we will fix this

David: * "together with the subjects generated by the subject map" - Similar to previous comment, changing "subjects"->"subject" I think removes ambiguity.
... improve wording
... * "The referenced columns of all term maps of a triples map (subject map, predicate maps, object maps, graph maps) MUST be column names that exist in the term map's logical table." - Per the previous definition of "column name" this implies that the values cannot be qualified. Is that the intent? So a column reference in a mapping cannot be "EMP.JOB", it must be "JOB"?

Richard: Does not reduce expressivity
... can add a short explanatory note

<cygri> ACTION: cygri to add note to R2RML section 6 to the effect that referenced columns are really unqualified column names and that's intentional [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action01]

<trackbot> Created ACTION-161 - Add note to R2RML section 6 to the effect that referenced columns are really unqualified column names and that's intentional [on Richard Cyganiak - due 2011-11-10].

David: Section 6.3 * "A predicate-object map is a function that creates predicate-object pairs from logical table rows." - Perhaps this would be clearer if it said "creates a predicate-object pair for each logical table row". This would avoid ambiguity about how many predicate-object pairs are created per logical table row.
... clarify wording as suggested
... Section 7.1 * "Occurrances of these properties..." - "Occurrances" -> "Occurrences" * "aaa rr:subject bbb." - Using aaa and bbb as sample values makes the table messy, IMO. Would "x" and "y" serve the same purpose but look cleaner?
... leave to edotor's discretion
... Section 7.2 * "the singleton set containing the value of rr:column" - For clarity, I think this should be changed to something like "... the value of _the rr:column property_". Otherwise it could be read as describing a set containing rr:column.

Richard/Seema reword for clarity

David: Section 7.3 * "Backslash characters MUST be escaped by doubling them with another backslash character." - This reads ambiguously to me. It could be read as saying to use two or three backslashes. This possibility for confusion is compounded by the Turtle backslash handling. For clarity it could be changed to something like "Backslash characters MUST be escaped by preceding them with another backslash character."

can we say "/" and "//"

David's comments are at: http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Oct/0022.html

<cygri> cygri: ok, adopt proposal to replace "doubling" with "preceding", add the actual character ("\") in brackets, and add an example for backslash doubling. same for the next sentence on curly brackets

David: * "If a template contains multiple pairs of unescaped curly braces, then adjacent pairs SHOULD be separated..." - I don't think the word "adjacent" is helpful here and it could lead to confusion. I think the word can simply be dropped.
... remove word "adjacent"

<cygri> ... then any pair should be separated from the next one by ...

Leave to editor's discretion

<cygri> cygri: i'm ok with changing to "... then any pair should be separated from the next one by ..."

David: * On first reading it is not clear how to produce an IRI from database columns without having the R2RML processor perform percent encoding. It seems to me to be worth adding a note mentioning that this can be done by creating a SQL query based logical table that includes an expression building up an IRI value from the database. This would avoid using a template and thus avoid the automatic percent encoding.

<cygri> ACTION: cygri to add note to 7.3 to explain that you cannot get an un-%-escaped iri from a template, but you can get it by using a view and a column [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action02]

<trackbot> Created ACTION-162 - Add note to 7.3 to explain that you cannot get an un-%-escaped iri from a template, but you can get it by using a view and a column [on Richard Cyganiak - due 2011-11-10].

David Section 7.6 * "...that does not have a specified langauge tag." - "langauge" -> "language"

scribe: typo

David: Section 7.7 * "A quoted and escaped data value is a SQL literal that can be used in a SQL query, such as.." - Seems to me that this needs a bit more definition. What does quoting mean? What does escaping mean?

Richard: Add reference to SQL production

<cygri> cygri: david, do you want a crisper definition, or more explanation of escaping in the spec?

<cygri> david: crisper definition. it's a rough spot in the spec

BREAK: Until 12:30 Eastern

<juansequeda> Ashok, is there a schedule somewhere? What issues are being addressed right now?

<juansequeda> It's a busy day with deadines, so I'm not sure when I would be able to call in, but I would like to be on the call when DM issues are being address

David's comments are at: http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Oct/0022.html

Juan, we are going thru David's comments

David: Section 8 * It would be good for this section to reference the place in section 11 that describes how triples are generated from the "joint SQL query". Otherwise this section appears to lead to a dead-end. It defines the joint SQL query but doesn't say what to do with it.
... add reference here to Section 11

Souri agrees

David: * "SELECT * FROM ({child-query}) AS tmp" - I would change this to just: child-query or: {child-query}
... just put the other query here

<Souri> family (h, w): TM_h and TM_w

<dmcneil> MenTriplesMap

<dmcneil> predicate: wife

<dmcneil> refObjectMap [

<dmcneil> parentTriplesMap WomenTriplesMap

<dmcneil> ]

<dmcneil> HUSBAND WIFE

<dmcneil> John Mary

David: We need a motivating example. It's a complex case
... * "SELECT * FROM ({child-query}) AS child, ({parent-query}) AS parent" - >From a SQL perspective, this feels wrong because it doesn't address the issue of column name collisions between the child and parent query. Is the thought that we can just ignore this because we describe in section 11 that the projected columns are split between child and parent?

Souri: Not sure what other databases do in this case

<scribe> scribenick: Ashok

<scribe> scribe: Ashok

Souri: SELECT {column-list} FROM ...

<Souri> right

We discused the fourth bullet

David: * I think it is worth noting in this section that the join condition between child and parent does not have to be a 1:1 relationship. It can be M:N.

Souri: Add note saying join could be M:N in general

David: Section 9.1

* "then the triples will share the same single blank node." - The word "single" seems un-needed and awkward in this sentence. I would remove it.

scribe: Fix the English here

Souri: how about "share that blank node"?

David: Section 10

* "and in hte" - change "hte" -> "the"

scribe: fix the typo

David: Section 11 * "The output dataset MUST NOT contain any other RDF triples or named graphs besides these." - I am curious to hear the thought process behind this text. I assume it is to have well defined behavior for interop and mapping portability? In our R2RML implementation, we intend to generate additional triples that provide metadata about the generated triples. So this text seems overly restrictive to us. I can see a few ways to address our need:

<Souri> at least those graphs containing those triples

<Souri> I am okay with either taking this out OR keeping it there and allowing implementations to provide a (session-level?) setting of flags to dictate whether additional triples may be included

<Zakim> ericP, you wanted to ask if we want to test against <http://example.org/Person/7> Person:age 33, 32, 34.

<ericP> http://www.w3.org/TR/sparql11-query/#extensionFunctions

<ericP> http://www.w3.org/TR/sparql11-query/#operatorExtensibility

Richard: I suggest adding a note

<Souri> +1

<Souri> +1 to ericP: "additional triples" would be more appropriate

<cygri> Richard: add a note in Section 11 along the line: "an implementation may offer access to a dataset (or multiple datasets) that contains additional triples in addition to those in the output dataset, such as provenance or inferred triples. the test cases however assume that there are no additional triples."

<ericP> +1

<cygri> ericP: take out the test case stuff?

<cygri> ashok: i'd take it out

<cygri> richard: ok let's remove the test case sentence

David: Section 11.1 * "Each generated triple is placed into one or more particular graphs" - I would strike the word "particular" from this sentence.

Agreement

* "has no object map (but a referencing object map)" - This sentence was hard to parse on first read. I think it would be clearer if it said "(but _it does have_ a referencing...)"

Agreement

* "has no referencing object map (but a normal object map)" - Same comment as previous comment.

Agreement

"Target graphs, a set of zero or more IRIs" - To my reading of the steps, the target graph is either a single IRI (i.e. rr:defaultGraph) or a set of IRIs. I don't think it is ever a set of size zero. Futhermore, I think the steps above need to be changed to produce singleton sets of rr:defaultGraph.

<cygri> cygri: in 11.1, there are three instances of "add the following triples to the output dataset", which mention "Target Graphs". check that the target graphs are always a set (in the case of rr:defaultGrpah); also one rr:defaultGraph is incorrectly capitalized

David: * "For each possible combination <s, p, o>" - For me this is a confusing way to describe the process. I don't think it is helpful to describe the subjects, predicates, and objects as sets and then take the cross-product of them. As far as I can tell this description only works in the document because the subjects, predicates, and objects are only ever singleton sets. Furthermore the steps are not written to produce sets. So I would change the descripti

<Souri> TM => S{PO}* is the way it is right now

<cygri> richard: allowing multiple s/p/o here was motivated by translation tables and by allowing multiple predicate maps on one p/o-map, both of these features were removed for LC

<cygri> ... i think we should revisit this question once we have resolved these features

David: * "adding triples to the output dataset" is defined as a term but the place where that term is referenced uses different text. This makes it hard to search the document for places whether the term is referenced. Therefore I propose using the same text in both the definition and the uses.

<cygri> david: also, sometimes it's "generate RDF term" (1) or "generate RDF terms" (multiple)

David: this editorial

Agreement

David: Section 11:2 * "A term map is a function that generates a set of RDF terms from a logical table row." - This statement is slightly different from the statement in section 7: "A term map is a function that generates an RDF term from a logical table row.". I think these two statements should match.

Richard: Requires cleanup after we resolve LC issues

David: * This section seems to have two different positions running through it: a) term maps produce a set of RDF terms (where the set is either empty or a singleton set?) b) term maps produce an RDF term or nothing It seems to me that we need to pick one or the other of these positions and make all the sentences consistent with that positions (both here and in earlier sections like section 7).
... same issue.

We can revisit later

* "if the term map references a NULL value" - Since we have a definition for "referenced columns" we should use that text here and make it a link. Then the language could be tightened to something like "if any of the term map's referenced columns have a NULL value"

<Souri> +1 to "referenced column" comment about Sec 11.2

Should say "if value of one of the referenced columns is NULL"

Agreement

<Souri> one or more?

<cygri> richard: yes, and "referenced columns" is a link to the relevant definition

David: Section B.2 * I found it a bit confusing to click on the property links (e.g. rr:child) and to be taken to another document (I expected the properties to be defined in the spec). Furthermore it is not clear to me (maybe I missed something?) whether this related document is normative: http://www.w3.org/ns/r2rml#

<cygri> ... yes, it should say, "one or more"

David: I would like Schema to be in the spec ... the pointer takes you to another (Schema) document

Richard: We could format differently to make it clear it's another document

<ericP> note that e.g. the XML spec has the formal representation (BNF, in this case) interspersed with the normative text, c.f. http://www.w3.org/TR/REC-xml/#sec-starttags

<Souri> I am ok with making the schema normative (old WD had this embedded in the spec)

<ericP> the bnf is recapitulated at the bottom

<ericP> (i've always found that a very readable and educational mixture)

<Souri> I like it to be separated, but also ok with making the separated doc normative

<mhausenblas> ACTION: Hausenblas to ask Ivan re how other SW groups handle(d) Schema normative (or not) [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action03]

<trackbot> Created ACTION-163 - Ask Ivan re how other SW groups handle(d) Schema normative (or not) [on Michael Hausenblas - due 2011-11-10].

David: * "Note that additional constraints not stated in this table might apply, and making a property forbidden or required in certain situations." - This doesn't seem to be a grammatically correct sentence to me.

Agree to fix the sentence

David: * With the addition of shortcuts, I think parts of this table are wrong. e.g. rr:objectMap is now an optional property, so the cardinality for it should be "0...1" rather than "1". I think this applies to several of the properties. But, actually it is quite useful from a user and implementor perspective to see a list of which _sets_ of properties are required. Meaning that either rr:object or rr:objectMap is required. One way to address this would be

<Souri> rr:table and rr:sql INSTEAD OF rr:tableName and rr:sqlQuery

<Souri> rr:logicalTableName and rr:logicalTableSql

<Souri> just stay with rr:table ... as a shortcut for rr:logicalTable [ rr:tableName ...]

<Souri> very confusing ...

<mhausenblas> Michael: I have to leave now, sorry - hope you still have a lot of fun - KUTGW!

OK... bye I will send you mail

<Souri> I am very much for allowing shortcut only for tableName b/c somebody writing a sql query is already dealing with complex spec

<cygri> ACTION: richard to create issue for the tableName sugar: 1. should we have it at all? 2. should it use different paragraphs? 3. should it apply just to tableName or also sqlQuery and sqlVersion? 4. the handling of this sugar in B.2 is non-obvious [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action04]

<trackbot> Created ACTION-164 - Create issue for the tableName sugar: 1. should we have it at all? 2. should it use different paragraphs? 3. should it apply just to tableName or also sqlQuery and sqlVersion? 4. the handling of this sugar in B.2 is non-obvious [on Richard Cyganiak - due 2011-11-10].

<cygri> ACTION-164

<cygri> ACTION-164?

<trackbot> ACTION-164 -- Richard Cyganiak to create issue for the tableName sugar: 1. should we have it at all? 2. should it use different paragraphs? 3. should it apply just to tableName or also sqlQuery and sqlVersion? 4. the handling of this sugar in B.2 is non-obvious -- due 2011-11-10 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/164

David: * With the addition of shortcuts, I think parts of this table are wrong. e.g. rr:objectMap is now an optional property, so the cardinality for it should be "0...1" rather than "1". I think this applies to several of the properties. But, actually it is quite useful from a user and implementor perspective to see a list of which _sets_ of properties are required. Meaning that either rr:object or rr:objectMap is required. One way to address this would be
... * "R2RML view" and "SQL base table or view" are two concepts that appear in the "Context" column but do not have corresponding classes in the schema. Both of these concepts are informal sub-types of LogicalTable. This seems like enough of a wrinkle that it warrants some additional explanation. I certainly had to spend several minutes tracing everything to find all the connections. Alternatively, why don't we create classes for these two concepts?

Richard: makes sense to add as subclass of LogicalTable

David: * The referenced schema for rr:sqlQuery, rr:sqlVersion, and rr:tableName do not specify a domain for these properties. Don't these all have a domain of LogicalTable?

Richard: Issue I just created should also mention this last point

Starting on the other LC issues

<dmcneil> http://www.w3.org/2001/sw/rdb2rdf/wiki/Last_Call

[1] Examples in 2.6 contain redundant rr:termType declaration Richard Cyganiak http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0000.html

RESOLUTION: Close issue by removing redundancy

[2] Note and examples on R2RML SQL identifier syntax disagree with normative text Richard Cyganiak http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0001.html

<Souri> commendable (b/c it was not "commentable")

RESOLUTION: Editors will fix this

Souri: Related to Issue 35
... syntax is a bit heavyweight

<Souri> "\"emp\""

[3] Cardinality of predicate maps in predicate-object maps Richard Cyganiak http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0004.html

<Souri> S{P(+)O}(*) shortcut: writing <P1 O> and <P2 O> with < {P1 P2} O> => I do not see any reason for complicating our language for this shortcut

<dmcneil> ashok & souri - i added a comment here like you requested: http://www.w3.org/2001/sw/rdb2rdf/track/issues/35

[4] Subject: RDB2RDF Last Call Working Draft transition announcement Bob Ferris http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0005.html

<ericP> off for a few... back soon

<cygri> ISSUE-62

<cygri> ISSUE-62?

<trackbot> ISSUE-62 -- Re-using public entity identifiers - external service call -- postponed

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/62

<Souri> Translation table is a deep topic, I'd say we revisit the issue and discuss during WG: however, note that if we add a translation scheme to the current LC then we do need another LC

[5] comments on working drafts Dominique Guardiola http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0006.html

<cygri> souri, might still be worth it imo.

Discussion postponed

<cygri> richard: regarding Bob's comment i think we should revisit the translation table issue, and reply with the outcome of that discussion; and also reply with our ISSUE-62 decision (postpone, potential R2RML 1.1 feature)

[5] comments on working drafts Dominique Guardiola http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0006.html

[7] R2RML: classes of triples map only constants Ahmed Bassiouni http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0015.html

<cygri> richard: see note at the beginning of http://www.w3.org/2001/sw/rdb2rdf/r2rml/#typing for a way of solving this

Richard: The note addresses his question
... perhaps note seems to be improved.

<Souri> richard, its ok with me to discuss this issue (Issue-62, Issue-66) again, but just wanted to remind us about the LC redo

<cygri> ACTION: richard to consider rephrasing the note that addresses http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0015.html [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action05]

<trackbot> Created ACTION-165 - Consider rephrasing the note that addresses http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0015.html [on Richard Cyganiak - due 2011-11-10].

[8] R2RML: comments on spec and test cases Toby Inkster http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0017.html

Editorial. Agrred to fix.

<cygri> Souri, rr:RefObjectMap still is a separate class, just the property rr:refObjectMap is gone.

<cygri> so only one of the three occurrences is an error

<Souri> ok thanks richard

[9] (Editorial) Last Call Comments on the Direct Mapping document Ivan Herman http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Oct/0000.html

Eric will look at these comments and report back. Seem to be all editorial.

[10] Editorial comments on R2RML Ivan Herman http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Oct/0001.html

R@RML editors will look at these editorial comments

<Souri> s/R@RML/R2RML/

Summary of Action Items

[NEW] ACTION: cygri to add note to 7.3 to explain that you cannot get an un-%-escaped iri from a template, but you can get it by using a view and a column [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action02]
[NEW] ACTION: cygri to add note to R2RML section 6 to the effect that referenced columns are really unqualified column names and that's intentional [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action01]
[NEW] ACTION: Hausenblas to ask Ivan re how other SW groups handle(d) Schema normative (or not) [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action03]
[NEW] ACTION: richard to consider rephrasing the note that addresses http://lists.w3.org/Archives/Public/public-rdb2rdf-comments/2011Sep/0015.html [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action05]
[NEW] ACTION: richard to create issue for the tableName sugar: 1. should we have it at all? 2. should it use different paragraphs? 3. should it apply just to tableName or also sqlQuery and sqlVersion? 4. the handling of this sugar in B.2 is non-obvious [recorded in http://www.w3.org/2011/11/03-RDB2RDF-minutes.html#action04]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.136 (CVS log)
$Date: 2011/11/03 19:22:16 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.136  of Date: 2011/05/12 12:01:43  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/cla/clarifies this./
Succeeded: s/commention/connection/
Succeeded: s/that/share that/
Succeeded: s/LV/LC/
FAILED: s/R@RML/R2RML/
Found ScribeNick: Ashok
Found Scribe: Ashok
Inferring ScribeNick: Ashok

WARNING: No "Topic:" lines found.


WARNING: Replacing list of attendees.
Old list: tpac dmcneil seema joerg mhausenblas cygri ericP
New list: OpenLink_Software


WARNING: Replacing list of attendees.
Old list: OpenLink_Software
New list: dmcneil ericP joerg tpac Souri mhausenblas seema MacTed juansequeda

Default Present: dmcneil, ericP, joerg, tpac, Souri, mhausenblas, seema, MacTed, juansequeda
Present: David Seema Michael Richard Ashok Eric
Got date from IRC log name: 03 Nov 2011
Guessing minutes URL: http://www.w3.org/2011/11/03-RDB2RDF-minutes.html
People with action items: cygri hausenblas richard

WARNING: No "Topic: ..." lines found!  
Resulting HTML may have an empty (invalid) <ol>...</ol>.

Explanation: "Topic: ..." lines are used to indicate the start of 
new discussion topics or agenda items, such as:
<dbooth> Topic: Review of Amy's report


[End of scribe.perl diagnostic output]