RDB2RDF Working Group Teleconference -- 12 Jul 2011

<trackbot> Date: 12 July 2011

<scribe> scribenick: ericP

<dmcneil> http://www.metalunderground.org/viewtopic.php?f=34&t=36815 ?

<mhausenblas> Μη μου τους κύκλους τάραττε

<dmcneil> so richard is the roman soldier

<cygri> today's agenda: http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Jul/0046.html

<mhausenblas> PROPOSAL: Accept the minutes of last meeting http://www.w3.org/2011/06/28-rdb2rdf-minutes.html

RESOLUTION: Accept the minutes of last meeting http://www.w3.org/2011/06/28-rdb2rdf-minutes.html

restructed R2RML spec

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

mhausenblas: kudos to cygri for work on r2rml

cygri: please read it
... rewrite goals:
... .. flatten table of contents
... .. remove duplicate defn's of the same property
... .. NOT change the language

<dmcneil> +q

cygri: for more info, read it

mhausenblas: would like a review

soeren: last time i put comments into a word doc

<mhausenblas> Ivan is on vacation

soeren: how should i annotate this time?

cygri: wiki or email, and just say e.g. section 6.3

soeren: i like to put the comments in context
... but will use wiki or email

dmcneil: i thought there were spec changes

<cygri> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Jul/0009.html

dmcneil: is there a list of the [non-editorial] changes?
... how many issues were brought forward?

cygri: http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Jul/0009 contains a list of changes i planned to make
... i did make all those changes

dmcneil: i thought that the template stuff had changed, but maybe it was just enumerated

cygri: enumerated, but didn't list all those in email
... i believe i captured or addressed all old issues

<dmcneil> -q

housekeeping

<Souri> we have a few "pending review" actions -- can we go over them and close (if possible)?

<mhausenblas> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Jun/0136.html

PROPOSAL: Resolve ISSUE-20, ISSUE-28, ISSUE-32 and ISSUE-44
... Resolve ISSUE-20, ISSUE-28, ISSUE-32 and ISSUE-44 per http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Jun/0136

<MacTed> +1 handle as in the email...

cygri: these were all dmcneil issues which dmcneil has agreed were addressed

RESOLUTION: Resolve ISSUE-20, ISSUE-28, ISSUE-32 and ISSUE-44 per http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Jun/0136

<cygri> ISSUE-34?

<trackbot> ISSUE-34 -- R2RML terminology: addressing vendor-specific names like "owner" -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/34

<cygri> PROPOSAL: Resolve ISSUE-34 by dropping rr:tableOwner, and stating that rr:tableName MAY be a schema-qualified name.

<dmcneil> +1

<Souri> rr:tableName "scott.emp"

<Souri> rr:tableName """"scott smith"."emp info""""

<cygri> ISSUE-35?

<trackbot> ISSUE-35 -- Case sensitivity of SQL identifiers -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/35

Souri: what will we do with a schema name "scott smith"?

<MacTed> SQL spec sets these containers... The Database >> Catalog >> Schema >> Tables/Views >> Rows/Columns

cygri: ISSUE-35 is in the same problem space (identifiers with quotes around them)

<dmcneil> +q

<Souri> rr:tableOwner "scott smith"

Souri: table ownder gives a nice way to label these things

<MacTed> qualified name is [[["Database".]"Catalog".]"Schema".]"Table"

Souri: we could make table owner optional

<MacTed> owner is often conflated with schema

<Souri> rr:tableQname

Souri: we could also call table name "schemaQualifiedName"

dmcneil: for me, if the mapping user wants to select a default schema, then they'd do that outside of r2rml, e.g. in connection properties.
... do others see it the same way?

cygri: there is text in the doc about connection properties outside the spec

MacTed: there are many reasons to do cross-schema joins
... trying to escape this stuff will brake a lot of users' and implementors' brains

<Souri> +1 to MacTed we need to reference something outside of default (connection) schema

cygri: we currently say that if you leave out rr:tableOwner, we'd use the default connection
... in a SQL query, you can explicitly name a schema and user

<Souri> I thought it was default schema issue was already decided that: if you do not specify a schema, it will automatically use the default (connection) schema

cygri: so this isn't specific to tableName

<dmcneil> +1 to proposal on table

<MacTed> -1 to dropping tableOwner

MacTed: addressing vendor-specific notions like owner is not address by dropping table owner

<MacTed> issue is "Owner" is vendor-specific, not SQL-standard

Souri: i thought that owner was not vendor-specific

<MacTed> SQL spec sets these containers... The Database >> Catalog >> Schema >> Tables/Views >> Rows/Columns

<MacTed> so use those terms...

<cygri> PROPOSAL: Resolve ISSUE-34 by dropping rr:tableOwner, and stating that rr:tableName MAY be a schema-qualified name.

<MacTed> columnTable

<MacTed> tableSchema

<MacTed> schemaCatalog

<MacTed> catalogDatabase

Souri: we need to explicitly specificy a schema

<MacTed> leave any of these out, and you're in the defaults

Souri: we could add it to the tableName but that's a different thing called schema-qualified name

MacTed: putting all of those identifiers in a single field brings up real escaping problems

<cygri> PROPOSAL: Resolve ISSUE-35 by allowing use of quotes around SQL column/table identifiers in the mapping language; requiring that authors match the quoting style that was used when creating the identifier; follow SQL rules for case sensitivity (quoted is case sensitive; unquoted is case insensitive)

MacTed: this problem doesn't address my issue with issue-34

<Zakim> ericP, you wanted to ask for specific use cases which aren't addressed by 34+35

<MacTed> PROPOSAL: Resolve ISSUE-34 by changing all vendor-specific terminology to spec-generic terminology; replacing rr:tableOwner with rr:tableSchema; and adding rr:columnTable rr:schemaCatalog rr:catalogDatabase to make complete identifier construction possible

ericP: what use cases can't be addressed at all by 34+35?

cygri: if you are writing a SQL query with case-sensitive table names, you need to put it in quotes
... if it's in a catalog, you need to use [[ FROM "catalog name"."table name" ]]

<Souri> example: select * from "scott smith"."emp info";

cygri: this is already the case
... my proposal for 34 is to put it in one property
... if someone wants to address a table in another schema, they'd reflexively prefix the schema name

<mhausenblas> PROPOSAL: Resolve ISSUE-34 by dropping rr:tableOwner, and stating that rr:objectName MAY be a schema-qualified name.

MacTed: propose to change tableName to objectName because tableName has a very specific interpretation

<MacTed> PROPOSAL: Resolve ISSUE-34 by dropping rr:tableOwner, and stating that rr:objectName MAY be a fully-qualified name.

<dmcneil> +1 to "fully-qualified"

<Souri> how would you express: "scott smith"."emp info" ? Is it: [] rr:tableName """"scott smith"."emp info""""? I do not see any major gain over: [] rr:tableName "scott smith" ; rr:tableOwner "emp info" .

<dmcneil> PROPOSAL: Resolve ISSUE-34 by dropping rr:tableOwner, and stating that rr:tableName MAY be a fully-qualified name.

<Zakim> ericP, you wanted to ask for use cases which are enhance by having schemas etc. distinguished

<dmcneil> not necessarily in the SQL queries

<dmcneil> in the R2RML property values

<cygri> quoting SQL spec: "Base tables and views are identified by <table name>s."

<Souri> figuring out whether a given name is schema qualified or not is a hassle: separating out as rr:tableOwner (or rr:tableSchema) and rr:tableName is much easier to process

<dmcneil> it's whether you qualify them in SQL or in RDF

Souri: we want to meet the Sept LC
... it's a hassle to crack owner, schema and table names
... we could add another prop for the qualified name

<Souri> PROPOSAL: no change or replace rr:tableOwner with rr:tableSchema

<Souri> PROPOSAL: no change or replace rr:tableOwner with rr:tableSchema, may be supplement with a new property rr:fullyQualifiedTableName

cygri: the real name for tableOwner is tableSchema
... but a schema can also have a catalog
... so you've pushed the problem on

<dmcneil> +q

cygri: so we need a third property, or four per MacTed's list
... so we're complciating the language to save us some parsing

dmcneil: if we want to separate them all, then we want to address them in RDF
... in which case want them to create (and exploit) a resource

<Souri> I have put in an alternate (extended) proposal

dmcneil: but to move on, i think we want to wrap it all into the tableName

+1 to dmcneil

<Souri> PROPOSAL: replace rr:tableOwner with rr:tableSchema *and* supplement with a new property rr:fullyQualifiedTableName

mhausenblas: we have three proposals:
... cygri's with 1 property
... MacTed's with 4
... Souri's with 3

+1 to cygri's proposal

<mhausenblas> ISSUE-35?

<trackbot> ISSUE-35 -- Case sensitivity of SQL identifiers -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/35

<mhausenblas> PROPOSAL: Resolve ISSUE-35 by allowing use of quotes around SQL column/table identifiers in the mapping language; requiring that authors match the quoting style that was used when creating the identifier; follow SQL rules for case sensitivity (quoted is case sensitive; unquoted is case insensitive)

second

<dmcneil> reading, trying to refresh my memory

ericP: aligns well with behavior of SQL implementations?

cygri: yes

<dmcneil> +q

cygri: whenever you'd make a ref to a table in SQL, you'd put the same text into the turtle string and get the same results

dmcneil: we've faced an issue around the default behavior of the DB
... we wanted to emulate the default for the DB

cygri: SQL 2008 defaults to case-insensitive unless names are in quotes

<cygri> This specification defines R2RML for databases that conform to Core SQL 2008, as defined in ISO/IEC 9075-1:2008 [SQL1] and ISO/IEC 9075-2:2008 [SQL2]. Processors and mappings may have to deviate from the R2RML specification in order to support databases that do not conform to this version of SQL.

cygri: if you want to run on something other than 2008, i don't see how we can do more to address it

<Souri> +1

<soeren> +1

dmcneil: i think this issue is still thorny, but support this resolution in the interest of progress

RESOLUTIONL Resolve ISSUE-35 by allowing use of quotes around SQL column/table identifiers in the mapping language; requiring that authors match the quoting style that was used when creating the identifier; follow SQL rules for case sensitivity (quoted is case sensitive; unquoted is case insensitive)

RESOLUTION: Resolve ISSUE-35 by allowing use of quotes around SQL column/table identifiers in the mapping language; requiring that authors match the quoting style that was used when creating the identifier; follow SQL rules for case sensitivity (quoted is case sensitive; unquoted is case insensitive)

<Seema> I am around

<Souri> I am around, but deadline bound a bit

mhausenblas: all, please view the issues in http://www.w3.org/2001/sw/rdb2rdf/track/issues/open

<mhausenblas> trackbot, end telecon

- DRAFT -

RDB2RDF Working Group Teleconference

12 Jul 2011

Attendees

Contents

restructed R2RML spec

housekeeping

Summary of Action Items

Scribe.perl diagnostic output