RDB2RDF Working Group Teleconference

<trackbot> Date: 19 April 2011

<mhausenblas> trackbot, start telecon

<trackbot> Meeting: RDB2RDF Working Group Teleconference

<trackbot> Date: 19 April 2011

<Marcelo> I am on irc, I will call in a few minutes

<ericP> scribenick: ericP

admin

<mhausenblas> PROPOSAL: Accept the minutes of last meeting http://www.w3.org/2011/04/12-rdb2rdf-minutes.html

<cygri> +1

<betehess> +1

<ivan> +1

APPROVED

RESOLUTION: Accept the minutes of last meeting http://www.w3.org/2011/04/12-rdb2rdf-minutes.html

F2F at TPAC

<ivan> 31 st October - 4 November, Santa Clara

<mhausenblas> see http://www.w3.org/2011/11/TPAC/

<juansequeda> semtech?

<juansequeda> in june?

<mhausenblas> juansequeda, it was not so much about the F2F but about TPAC ;)

cygri: there's a potential semi-conflict with RDF WG

ivan: we should see RDF poll results tomorrow

<mhausenblas> Michael: I agree that we should plan a F2F independent of TPAC

ivan: chicken and egg with RDF
... do we *need* a f2f?

mhausenblas: f2f's productive if folks are prepared to close issues
... if we can get editors to resolve stuff, we might need two days

cygri: when in LC?
... charter says "T+18[months]"; when did we start?
... [group arithmetic break]
... we should already be at LC

<mhausenblas> ACTION: Hausenblas figure out runtime and if charter extension is necessary [recorded in http://www.w3.org/2011/04/19-rdb2rdf-minutes.html#action01]

<trackbot> Created ACTION-117 - Figure out runtime and if charter extension is necessary [on Michael Hausenblas - due 2011-04-26].

cygri: first meetings in Oct 2009, so T+18 is Apr 2011
... if we don't have LC by Nov, something is wrong

ivan: if we get wide review by don't get many comments, we don't need an F2F
... if we get lots of comments, an F2F in Nov is appropriate
... we should formulate amoungst ourselves a proposal for the WG.
... TPAC is the week after ISWC

mhausenblas: we also have EU project calls/submissiond deadlines around Nov

juansequeda: how about at ISWC?
... Semtech? probably too early

ivan: yeah, too early, and i'm overloaded

<juansequeda> ericP, in Santa Clara, CA

Souri: SPARQL f2f [virtual video circuit] was very effective

ivan: won't have equipment at ISWC or TPAC

you can relocate some of the people all of the time and you can relocate all of the people some of the time, but you can't relocate all of the people all of the time

(unless you're Stalin)

ivan: RDFa F2F useful to coincide with HTML meeting

<mhausenblas> ACTION: Hausenblas to put the F2F meeting onto 2011-04-26 agenda [recorded in http://www.w3.org/2011/04/19-rdb2rdf-minutes.html#action02]

<trackbot> Created ACTION-118 - Put the F2F meeting onto 2011-04-26 agenda [on Michael Hausenblas - due 2011-04-26].

mhausenblas: right, makes sense in a wider context with e.g. RDF meeting

Outstanding Issues

<mhausenblas> http://www.w3.org/2001/sw/rdb2rdf/track/issues/open

<mhausenblas> scribenick: nunolopes

<mhausenblas> ISSUE-18?

<trackbot> ISSUE-18 -- Allow SQL queries to be reused in a mapping -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/18

Souri: multiple triples map could use the same SQL query

… we could associate a uri with a SQL string and then use it in multiple triplesMap

… we will come up with a proposal for that

<mhausenblas> ACTION: Souri to propose a solution for ISSUE-18 (SQL string/multiple triplesMap) - new property [recorded in http://www.w3.org/2011/04/19-rdb2rdf-minutes.html#action03]

<trackbot> Created ACTION-119 - Propose a solution for ISSUE-18 (SQL string/multiple triplesMap) - new property [on Souripriya Das - due 2011-04-26].

<mhausenblas> ISSUE-19?

<trackbot> ISSUE-19 -- Reuse a SQL query as a sub-query -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/19

Souri: this issue might cause problems with the constructions of select clauses

… are we going to use it that much that justifies the complexity?

mhausenblas: in this case we should pose the question back to david that raised the issue

David: it is very usefull, but I agree it is too complicated to get into the spec at this time

cygri: I'd be curiours to know how you handled this

david: we allow to give a subquery a name, and in a nother query you can use that name

cygri: in the SQL query you can refer by name a query otherwise defined in the query?

… implicitly defining a view in the mapping file and using it in the queries

david: it's very usefull to avoid duplicating code

cygri: you said it's too complicated but maybe we should propose something to assing a uri to a subquery?

… it seems that in terms of the language it wont' be too complicated

… altough might make implementations harder

mhausenblas: is someone willing to write up something on this?

… otherwise we postpone it

cygri: can we revisit this after issue 18 has been resolved?

<Souri> +1 to postponing Issue-19

<mhausenblas> PROPOSAL: Postpone ISSUE-19 and make ISSUE-18 block it

<Souri> +1

<cygri> +1

RESOLUTION: accepted to postpone ISSUE-19 and make ISSUE-18 a blocker

<mhausenblas> ISSUE-25?

<trackbot> ISSUE-25 -- Including direct mapping constructs in R2RML mappings -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/25

david: this issue is about allowing to use direct mapping constructs in R2RML mappings

juansequeda: should a user start with a blank mapping file then?

david: we have an equivalnet to the direct mapping, but we find that custumers don't use it

… they start with a domain ontology

…. you can start with the direct mapping and then edit it

… instead of making the constructs of the direct mapping in the R2RML

… so the engine doesn't need to know the direct mapping

… I'm objecting to having only a partial R2RML and the rest coming from the direct maping

juansequeda: I would object to that too

… is that mentioned somewhere?

Souri: what I recall is that a mapping should be specified in an incremental way

… start with the direct mapping and move from there

…. but then R2RML would have to process also the parts that are direct mapping

… I understand that this puts more burden on the engine

… as opposed to a more manual way of doing this

… i'm ok with the proposed approach at the moment

… when R2RML becomes more stable we can think of this hibrid approach

<MacTed> (how far to future has this group been extended, so I can adjust my calendar reminders?)

mhausenblas: is this a question about coupling between direct mapping and R2RML?

<betehess> is the Direct Mapping expressed in term of R2RML?

david: it's fine if the direct mapping is completely specified

… the problem is if it is not and we assume it should be complemented by constrcuts from the direct mapping

<betehess> no

david: leaving the spec as is satisfies my requeirement

Souri: yes, this feature is not reflected in the document

cygri: issue 25 was actually proposed by souren

ivan: yes, souren was in favour of the approach that david is now objecting

… so we should let soeren respond before closing this issue

<Souri> +1 to waiting for Soeren

cygri: I'd propose david writes an email to the list about this and would give soeren a chance to respond

<mhausenblas> ACTION: David to write a proposal to the WG list (close ISSUE-25) giving Soeren the opportunity to react [recorded in http://www.w3.org/2011/04/19-rdb2rdf-minutes.html#action04]

<trackbot> Created ACTION-120 - Write a proposal to the WG list (close ISSUE-25) giving Soeren the opportunity to react [on David McNeil - due 2011-04-26].

<mhausenblas> ISSUE-29?

<trackbot> ISSUE-29 -- Require blank node and IRI identifier expressions to produce strings -- open

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/issues/29

<Souri> didn't Eric work on it?

david: if we can have an expression that evaluates to a bnode or uri , should evaluate to a string to I can use string operations on these

Souri: I think eric suggested we cast it to string

david: I have some reservations about relying on database specific behaviour for this

… for instance formatting dates, etc

… doesn't seem like the best solution for me

Souri: maybe we should wait for eric on this

mhausenblas: can you take an action to write an email? to have a an action associated with the issue

<cygri> http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2011Mar/0090.html

Souri: I think we still need a discussion on this

… I can take an action to follow up on this

Souri: we want the R2RML processor to enforce the type of the expression?

… we may not even know the type beforehand

<mhausenblas> ACTION: Souri to follow-up on Eric's post regarding ISSUE-29 concerning string casts [recorded in http://www.w3.org/2011/04/19-rdb2rdf-minutes.html#action05]

<trackbot> Created ACTION-121 - Follow-up on Eric's post regarding ISSUE-29 concerning string casts [on Souripriya Das - due 2011-04-26].

Souri: so we could check the type of the columns and possible trhow an error..

<mhausenblas> ACTION-121?

<trackbot> ACTION-121 -- Souripriya Das to follow-up on Eric's post regarding ISSUE-29 concerning string casts -- due 2011-04-26 -- OPEN

<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/121

cygri: personally I don't like this, from a user's point of view I excepct that if the implemenetation complains this should be a string and I have to explicitly cast it as a string

<mhausenblas> Michael: Souri, if you look into the ACTION-121, there is the pointer to Eric's post

… the user might think it's a job for the engine

david: there's formatting decisions to be made

cygri: my feeling is that in the SQL spec specficies a canonical way to represent a string

… I think eric was supposed to look into this

mhausenblas: let's continue this on the list, we're out of time

<mhausenblas> Michael: Next week Ashok will chair, continuing with ISSUE-32

<mhausenblas> [meeting adjourned]

<mhausenblas> trackbot, end telecon