Re: Proposal for ISSUE-12, string literals

Re: plain literal other canonicalization, here are my priorities in
descending order of urgency:

  1. no MAY wording - I want SPARQL, N3, and every other system which
  is allowed to work over the unentailed graph to be working with the
  same graph. 

  2. pick exactly one of "abc"^^xsd:string and "abc" (preferably "abc")
  to be allowed in the abstract syntax.

  3. offer a datatype, e.g. xsd:string, to express that in e.g. range
  constraints. This serves as advice to RIF and SPARQL when tests are
  made on the datatype and to APIs when constructing terms, e.g.
    setDataType(createLiteral("abc"), xsd:string)

  4. don't unify "abc" and "abc"@en (does "chat"@en == "chat"@fr?).

  5. don't damn datatype extensibility by demanding canonicalization
  of all literals.

  6. either DO or DON'T (no MAY) canonicalize the beloved datatypes in
  XSD. (Again, this serves as advice to e.g. parsers and APIs).


Received on Friday, 13 May 2011 14:36:21 UTC