See also: IRC log
<Ashok> scribenick: dmcneil
minutes from last week http://www.w3.org/2011/09/27-RDB2RDF-minutes.html
accept minutes from last week?
RESOLUTION: accept minutes from last week
ashok would like to address two items that were not on the agenda 1) comments we have received so far
2) if eric joins us, talk about ISSUE-69
souri, what have you found in the last call comments?
souri replied to the commenters answering some immediate questions, and letting them know we will follow up with the other questions
any other reactions to last call comments?
ivan: we need a very precise record of all last call comments, and references to the answers, and indication of whether the response was accepted by the commenter
<Souri> That's how we did it in the SPARQL 1.0 working group ... it worked out well as far as I remember
this is needed to demonstrate to the w3c that we addressed all comments, in particular those from outside of the working group
this could be done in a wiki page or in the issue tracker
<Zakim> cygri, you wanted to ask for an example how other WGs did it
ivan could try to provide a link to example wiki pages from other working groups
ashok: with an example we could
start a page for the group to use
... what did sparql working group do?
eric: for that working group we left the questions and the handling of them in the mailing list
most comments were addressed by pointing commenter to a section of the spec
some involved modification of the spec
<ivan> http://www.w3.org/2007/OWL/wiki/Responses_to_Comments#First_Round
they used the subject line to indicate when a comment was "satsified"
ivan: this is a link to how the OWL group handled last call comments
this was easy to do, the work could be distributed, members just had to know a little bit about wiki
ashok: a bit biased towards wiki
ivan: the RDFa working group created a new product in Tracker for the Last Call
this allowed mailing list items to be tied into Tracker items
eric: mailing list was all that existed at the time for SPARQL working group
ashok: ivan, michael, and ashok can talk about it
next agenda item, ACTION-159
ACTION-159
<ivan> ACTION-159?
<trackbot> ACTION-159 -- Boris Villazón-Terrazas to create RDFa representation of R2RML vocab -- due 2011-09-27 -- OPEN
<trackbot> http://www.w3.org/2001/sw/rdb2rdf/track/actions/159
<boris> http://www.w3.org/ns/r2rml
boris has started on it and is still working on it, would like the working group to review it
ivan: it is 99% done
next agenda item, the test cases
<boris> http://www.w3.org/2001/sw/rdb2rdf/test-cases/TCOverview.html
<Ashok> http://www.w3.org/2001/sw/rdb2rdf/test-cases/TCOverview.html
<boris> http://www.w3.org/2001/sw/rdb2rdf/wiki/R2RML_TC
ISSUE-69?
<trackbot> Sorry... I don't know anything about this channel
http://www.w3.org/2001/sw/rdb2rdf/track/issues/69
ashok: can you summarize the differences (eric & richard)
eric: do we give people a recipe or a behavior specification
e.g. how to represent a decimal in an RDF graph
define a maximum size for the purposes of interop
e.g. dates vary in size, depending how far in the future or past needs to be referenced more or less digits are needed
for dates SQL only allows from 1 to 9999
most implementations exceed that, but they are outside of the spec
so what is the minimum behavior to expect from an implementation
cygri: it is simpler that this, we will define the mapping from a SQL-2008 database to RDF
but SQL-2008 leaves the number of supported decimal digits open to implementors
e.g. Oracle allows up to 38 digits
XSD specs only allows for up to 18 digits
according to what eric is proposing, Oracle would have to throw away 20 digits
this seems bizarre to me
eric: they don't have to throw it away, it is just this is not defined in the direct mapping
this is outside of the spec, and interop may not be possible
cygri maintains that the spec mandates throwing away digits to be compliant
<MacTed> this is handled in ODBC and similar by implicit transformations from (for instance) a numeric, e.g., INT(38), to a string, e.g., CHAR(38) -- with the idea generally to preserving the value, but not necessarily the datatype
cygri: it is very common for relational databases and RDF implementations to allow more than 18 digits
eric: I need to specify the minimum bar that implementor must meet so that we can write test cases
ashok: do others have opinions?
ted: ODBC handles this by allowing implicit data transformations, for example from a number with 38 digits to a string with 38 characters
this is spelled out to a degree in ODBC, but it does not specify everything
dmcneil: are we just talking about the direct mapping?
<cygri> +1 to ashok
<Souri> +1 to putting it in both
ashok: I thought each spec should have a section on this
so we are talking about both specs
<ericP> http://www.w3.org/TR/2011/WD-r2rml-20110920/#datatype-conversions
<ericP> http://www.w3.org/TR/2011/WD-rdb-direct-mapping-20110920/#defn-literal_map
eric: what numbers can be used in test cases
cygri: the maximum size of numbers is specified in the SQL specs
for decimal, it leaves the upper range open
eric: what is the minimum number of digits that a SQL implementation must support for decimal numbers
cygri: why does this matter?
eric: it affects what we can test against
this tells the user what size of number they can plan on using with R2RML
ted: what happens when the database exceeds the XSD size?
ashok: the question is: do we have to worry about it?
ted: we could say that this behavior is undefined
when we try to test edge-cases this becomes relevant
ashok: possible position, we write that if your datatypes are within the SQL standard and the XML schema standard then this is the behavior, if they are bigger then the behavior is not defined, and we don't write test cases for these
<ericP> Ashok: "if your datatypes are inside of SQL *AND* XSD, this is the behavior. if not, it's undefined and we won't write test cases [outside of the intersection of the intersection of XSD *AND* SQL]"
ted: so we would remove the part of the direct mapping spec that refers to discarding digits
eric can take a shot at writing this up for the spec
cygri doesn't see a case for changing the approach of the R2RML spec, but could add wording saying behavior is undefined if the data is outside of the XSD datatype size restrictions
eric: as a user I would like to know what the minimum implementation requirement is
long discussion that lead to a discussion of vendor specific types
ashok: my inclination is to say less about what happens when the data is outside of the spec
<Souri> +1 to Ashok's proposed text
wiki page says some implementations will be available in early 2011
but we can talk about this next week
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/implemenation/implementation/ Found ScribeNick: dmcneil Inferring Scribes: dmcneil Default Present: +1.314.394.aaaa, Ashok_Malhotra, Ivan, +49.133.6.aabb, boris, +3539149aacc, +575737aadd, juansequeda, nunolopes, +1.562.714.aaee, cygri, +1.603.897.aaff, Souri, +1.603.897.aagg, seema, EricP, +1.781.273.aahh, MacTed Present: David Ivan Boris Nuno Richard Juan Marcelo Souri Seema Eric Ted Regrets: Michael Percy Got date from IRC log name: 04 Oct 2011 Guessing minutes URL: http://www.w3.org/2011/10/04-RDB2RDF-minutes.html People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]