Re: Untyped Literal Semantics

> Third, I suggest we are choosing between one of three options:
> 
>    o untyped literals are self denoting
>    o untyped literals denote a value from the value space of some datatype
>    o we can't decide

I believe there is another disjunct decision that the WG should make, as suggested
by Jeremy, which would insulate most of the documents which must deal
with the concrete and abstract syntax from the above decision; and that
is to make inline literals syntactically (not semantically) untidy in the abstract
syntax (and NTriples), and assign to them a unique systemID such as used to 
name bnodes.

Thus

   <rdf:Description rdf:about="#Jenny">
      <age>10</age>
   </rdf:Description>

would be represented in the abstract syntax as

   Jenny age (_:x, "10") .

Adopting this form of syntactically untidy representation for non-typed
literals will also facilitate developers who opt for untidy semantics, even if 
we opt to not decide ourselves on the semantics of non-typed literals.

Note that making untyped literals syntactically untidy does not force
any application from using an internally tidy representation in the interests
of efficiency, etc.

I would like to see this addressed by the WG before or in parallel to the
separate and important issue of semantics and, if so adopted, reflected in the
editorial changes being made to reflect Part 1 of the restructured
datatyping specification in the various documents.

Regards,

Patrick

Received on Monday, 9 September 2002 06:54:49 UTC