<trackbot> Date: 03 April 2012

<mhausenblas> scribenick: nunolopes

Admin

<mhausenblas> PROPOSAL: Accept the minutes of last meeting http://www.w3.org/2012/03/13-RDB2RDF-minutes.html

<boris> +1

RESOLUTION: Accepted the minutes of last meeting http://www.w3.org/2012/03/13-RDB2RDF-minutes.html

Implementations

Implementation report/EARL

Implementations report/EARL

mhausenblas: Any news on implementations?

juansequeda: we'll have ultrawrap by the end of the month

… doing final checks

cygri: D2RQ has a preliminary implementation of the DM, nothing for R2RML

… we haven't run the test-suite but we should be complete except for 2 issues

… we can talk about that once we run the tests

… we hope to make a preview release of this

… and R2RMl should be next on the list

… in 6 weeks time

ivan: it will not be done before the end of CR

cygri: no ..

nunolopes: report on XSPARQL

cygri: the binary test case might be an issue

<Zakim> ericP, you wanted to say that SWObjects has a dm-materialize which passes the DM tests in MySQL and Postgres -- futzing around with Oracle and SQL Server

ericP: I have an implementation that passes the tests with mysql and psql and trying oracle and ms server

juansequeda: I'm trying over oracle, sql server and db2

ericP: the test harness I'm using actually creates the database, maybe this will be useful for the group

… if people are interested

cygri: for us, we haven't looked at test harness software

… we'll next look at what people did and more options to pick better

… the easier to run the better

ericP: there are also the native libraries

cygri: and JDBC drivers

ericP: in case we have circular dependency on foreign keys, it's a pain to add in the sql script and to delete the tables

… in my tests I don't tell the database there isn't a foreign key

… only when it is materialised

ivan: we do have 3 implementations of DM

… essentially complete by the end of CR, which is fine

… what about R2RML

juansequeda: ultrawrap does an implementation

nunolopes: I have one but you might be interested in more production ready ones

ivan: I was hoping for oracle and/or revelytix

<Seema> we can't say anything for now

… if we don't have that done by end of april we can't exit CR

Seema: we can't say anything yet

MacTed: any news on openlink?
... not at the moment

mhausenblas: next telecon will probably be april 24th and that will be the cut-off date

juansequeda: do the implementations need to pass all the test cases?

ivan: every test (or feature at least) needs to be passed by at least 2 implementations

<boris> yes

cygri: we don't have anyone from revelytix on the call but the literature says they cover R2RML

… there might be hope from that side

<mhausenblas> ACTION: Hausenblas to ping Revelytix re R2RML implementation status [recorded in http://www.w3.org/2012/04/03-RDB2RDF-minutes.html#action01]

<trackbot> Created ACTION-203 - Ping Revelytix re R2RML implementation status [on Michael Hausenblas - due 2012-04-10].

cygri: what happens at the end of the month?

ivan: if everything is as planned we need an implementation report

… check if the tests reveled any major issues in the spec

… if not we can start a proposed recommendation phase

cygri: assuming some feature doens't have 2 implementations, what happens?

ivan: we can wait for implementations or remove it from the document and go back into last call phase
... even if D2R doesn't have an implementation might be important to document what you have

cygri: in the implementation of the DM we've identified some issues,

… which should be fixed if we had no time dependencies

… since we are not in working draft maybe we can implement it anyway

… what's the protocol to deal with this at this point?

ivan: it's difficult to say in abstract terms

… RDFa 1.1 is in the same phase and the implementations revealed inconsistencies in the spec text

… what you say is to change the core spec, which would mean to go back to design phase

… if is something that is not perfect we can come back to it for a new version

… but would prefer to not go into another technical discussion at this stage

mhausenblas: we can mark it for future work

cygri: there is a difference between new features and issues that affect implementability

… might be so hard that we can say we don't do it

… for example, the handling of binary types

… when implementing, we think it should say xsd:hexbinary

mhausenblas: that's the kind of thing we should fix

cygri: that would be the new technical discussion ivan mentions no?

ivan: maybe, is the experience shared with the other implementors?

… would any change there change the results and implementations of the others?

… at first glance it doesn't seem like a big change

mhausenblas: since it's not a new feature it makes sense to discuss it based on implementations

… and find out what were the issues

… implementation dependent or related to the spec

… only if two implementors have different opinions we need to discuss it

cygri: the base64binary might be more useful, but of marginal interest

… more of a corner case that should be easy to implement

… and not have implementors waste too much time on it

ivan: let's hear from the other implementors

cygri: to make it clear dumping the database is not a problem

… sparqling the results might be a problem

ericP: when I implemented this I simply used an off-the-shelf algorithm

…. because I'm only materialising it's easy

cygri: the problem is if you want to push the join to the database

… the conversion between the types is much simpler for hexBnary than base64

cygri: I can document this via email

mhausenblas: maybe it's worth to document it on the wiki as well

<mhausenblas> ACTION: Hausenblas to create a new Wiki page for gathering implementation experiences [recorded in http://www.w3.org/2012/04/03-RDB2RDF-minutes.html#action02]

<trackbot> Created ACTION-204 - Create a new Wiki page for gathering implementation experiences [on Michael Hausenblas - due 2012-04-10].

TC and Test Harness

boris: I created some java code to compare the official results from results from an implementation

… and generates an EARL file

<boris> http://www.w3.org/2001/sw/rdb2rdf/wiki/Submitting_Test_Results

… describing which test cases are passed

… I've updated the webpage

<boris> http://mccarthy.dia.fi.upm.es/rdb2rdf/tc/th/earl.ttl

… implementors just need to provide the tool name and some config and run the script

… need to do the same for R2RMl

ivan: once all the earl files are available we need to process the files and generate the report

boris: yes, it's in my todo list

mhausenblas: thanks, boris, great work!

boris: thanks for everyone who check the testcases

<boris> http://www.w3.org/2001/sw/rdb2rdf/test-cases/#D016-1table1primarykey10columns3rowsSQLdatatypes

cygri: is the source for the TH available?

<boris> https://github.com/boricles/rdb2rdf-th

boris: yes, it will be in github

<joerg> uh

<mhausenblas> [meeting adjourned]

<joerg> thats bad

<mhausenblas> Next meeting is scheduled for 24 April due to WWW2012

<joerg> ok

<mhausenblas> trackbot, end telecon

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/beter/better/
Succeeded: s/open link/openlink/
Found ScribeNick: nunolopes
Inferring Scribes: nunolopes
Default Present: Ivan, boris, juansequeda, MacTed, mhausenblas, nunolopes, Seema, cygri, EricP
Present: Ivan boris juansequeda MacTed mhausenblas nunolopes Seema cygri EricP Boris Juan Ted Michael Nuno Richard Eric
Regrets: Souri Dave
Agenda: http://lists.w3.org/Archives/Public/public-rdb2rdf-wg/2012Apr/0001.html
Found Date: 03 Apr 2012
Guessing minutes URL: http://www.w3.org/2012/04/03-RDB2RDF-minutes.html
People with action items: hausenblas

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.