Re: R2RML draft - new introduction

* Richard Cyganiak <richard@cyganiak.de> [2010-10-15 13:21+0100]
> Juan,
> 
> On 14 Oct 2010, at 20:09, Juan Sequeda wrote:
> >>The output graph has to contain information that's in the
> >>database tables.
> >>So the tables themselves have to be part of the input.
> >>
> >
> >A relational schema consists of tables.
> 
> You are right (and so is Ashok), and I was wrong.
> 
> In my head, a “schema” is something that describes the structure of
> some data, and is distinct from the “actual” data. That's how the
> term is used in “XML Schema” and “RDF Schema”.
> 
> But having started to read the actual definitions of terms in the
> SQL spec, I now realize that a “SQL-schema” is indeed not just a
> collection of *table definitions*, but a collection of *tables* (and
> other stuff), consisting of columns and rows. So a SQL-schema
> includes the actual data.

Huh, popular interpretation (as measured by g:"sql schema") seems to
follow your interpretation. Given this intuition mismatch, could you
cite the bit of the SQL spec which appears to align more with what
dbpedia calls Oracle's "Schmea object"
  http://en.wikipedia.org/wiki/Schema_object
than with "Database schema"
  http://en.wikipedia.org/wiki/Database_schema
?


> I will update the draft to reflect that.
> 
> Sorry for the noise caused by my confusion. A bit embarrassing.

I guess we've all learned a little lesson about the consistency of terms.


> Best,
> Richard

-- 
-ericP

Received on Friday, 15 October 2010 13:57:12 UTC