RDB2RDF Working Group Teleconference -- 22 Nov 2011

<trackbot> Date: 22 November 2011

<Ashok> meeting: RDB2RDF

<ericP> scribenick: joerg

<ericP> scribenick: ericP

example scribing session

ashok: write the name followed by a ':'

<scribe> ... continued text prefaced by "..."

<joerg> ashok: introduced joerg to scrbing

joerg: all right, i'll do it

<Ashok> scribenick: Joerg

<ericP> scribenick: joerg

<Ashok> scribe: Joerg

<Ashok> PROPOSAL: Accept the minutes of the last meeting http://www.w3.org/2011/11/15-RDB2RDF-minutes.html

<cygri> +1

ashok: any objections?

<ericP> RESOLVED: Accept the minutes of the last meeting http://www.w3.org/2011/11/15-RDB2RDF-minutes.html

thanks

cr process

<cygri> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Nov/0070.html

<ivan> links to cr issues

s/F2F meeting/the last meeting/

<Ashok> Ivan: Changes since LC must be documented

<Ashok> Ivan: Resolution of LC comments must be documented

Ivan: there has to be a transition call
... sent some links around in the mail

<Ashok> Ivan: There has to be a call with the Directors. The chairs and staff contact must be present. Editors may be prsent to answers questions

<cygri> cygri: does the report on changes have to include editorial stuff, beautifying etc?

<cygri> ivan: well, no one is *really* interested in that

Ivan: stresses importance of chasing last call comments

<cygri> cygri: should we chase up commenters who didn't acknowledge our response?

<cygri> ivan: would be good, but of course there will always be some who don't reply

datatype mapping issue

data type mappings

<cygri> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Nov/0072.html

ashok: eric and richard are both here to discuss the topic

<cygri> and also: http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Nov/0048.html

richard: first links is the reaction to the break down of the problem
... canonical form is opposed beacuase of implementation cost
... sql recipes are not welcomed as normative definition
... new proposal: both canonical and not necessarily canonical form
... for literals-> not necessarily canoncial, IRIs -> canonical
... r2rml leaves iri creation to implementors , sql recipies get imformative

<dmcneil> +q

richard: dm has to be stricter
... canonical literals can be harder to read

<cygri> david: what happens in the case where you turn an integer column in a language-tagged literal?

<cygri> ... do you canonicalize there?

<cygri> cygri: it's kind of the same case as with IRIs

<cygri> ... and there are more

<dmcneil> e.g. blank node identifiers

<dmcneil> +q

<cygri> david: how do we ensure that INT 5 and BIGINT 5 are translated to the same thing?

<cygri> ... the spec needs to nail that down

ashok: what about the implementation costs

richard: don't bother the users to much which problems most likely not to be happening to often
... but reagrding to bigint 5 and int 5 problem, most importantly, the result should be conistent

<ericP> joerg: do we require SQL or XSD canonicalized form?

<Ashok> In R2RML you can choose not to canonicalize

richard: it is a should, so the implementer is free to stick to it.
... and most of the work is already done, by having examined the differences
... so the difference is not so big

<ericP> joerg: from an implementor's perspective who tries to do as much as possible in the database, XSD could be a costly choice

richard: query execution can be done completely in sql space

ashok: can somebody sum the discussion a bit up, Richard?

<ericP> +1 to 0072 plus amendments

richard: email of today seems consensual

<cygri> PROPOSAL: resolve ISSUE-69 by adopting proposal at http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Nov/0072.html with amendment from david regarding consistent translation of different SQL types that produce same XSD type

<MacTed> is david's amendment to-be-written? or can it be linked?

<dmcneil> plus the amendment to say it applies to plain literals as well?

<dmcneil> and words about blank node identifiers?

<cygri> PROPOSAL: resolve ISSUE-69 by adopting proposal at http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Nov/0072.html with amendments from david regarding (i)consistent translation of different SQL types that produce same XSD type and (ii) applying the design for IRI templates wherever "natural RDF lexical form" is currently used

<dmcneil> +1

<MacTed> +1

<ivan> +1

<ericP> +1

RESOLUTION: resolve ISSUE-69 by adopting proposal at http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Nov/0072.html with amendments from david regarding (i)consistent translation of different SQL types that produce same XSD type and (ii) applying the design for IRI templates wherever "natural RDF lexical form" is currently used

<Ashok> 914-271-6477

<cygri> ACTION: cygri to implement ISSUE-69 resolution [recorded in http://www.w3.org/2011/11/22-RDB2RDF-minutes.html#action01]

<trackbot> Created ACTION-172 - Implement ISSUE-69 resolution [on Richard Cyganiak - due 2011-11-29].

<cygri> ACTION: ericP to discuss editorial approach for ISSUE-69 with Richard [recorded in http://www.w3.org/2011/11/22-RDB2RDF-minutes.html#action02]

<trackbot> Created ACTION-173 - Discuss editorial approach for ISSUE-69 with Richard [on Eric Prud'hommeaux - due 2011-11-29].

<cygri> ericP, if you have any requests/ideas/proposals for how to structure R2RML Section 10, please share

- DRAFT -

RDB2RDF Working Group Teleconference

22 Nov 2011

Attendees

Contents

example scribing session

cr process

datatype mapping issue

data type mappings

Summary of Action Items

Scribe.perl diagnostic output