Re: [DTB] Action 681 completed

p.s.: I also removed the now obsolete predicate hasDatatype, which was 
the "predecessor" of

   isLiteralOfType
and
   isLiteralNotOfType

and added the respective examples and Editor's notes now under 
isLiteralOfType.

Two more questions open:

1)  I thought whether we also need:

    IsLiteral, IsNotLiteral

  While the former can be  easily emulated by: idLiteralOfType (l ?X ), 
just leaving the variable free, especially the latter might be useful? 
Opinions?

2) Naming convention... I know we had agreed on isLiteralOfType and 
isLiteralNotOfType in the teleconf., but now, in the light of drafting

literal-equal and literal-not-equal

I ask myself which naming convention to stick to:

  CamelCase or dash-separated ?

best,
Axel



Jos de Bruijn wrote:
> <snip/>
> 
>>>> Note (also an editor's note in the document):
>>>>  I assumed the second argument of isLiteralOfType to be a rif:iri at the
>>>> moment. As we defined a datatype identifier just as a unicode string
>>>> representing an IRI in the definition of symbol spaces, it might be
>>>> better to restrict the domain of the second argument to strings, yes?
>>> I disagree. A rif:iri constant can denote an actual datatype, so you can
>>> speak about actual datatypes when speaking about the types of literals.
>> This is what we say so far:
>> http://www.w3.org/2005/rules/wiki/DTB#Symbol_Spaces
>>
>> "The identifier of a symbol space is a sequence of Unicode characters
>> that form an absolute IRI."
>>
>> It is not an IRI constant, although the current definitions of
>> isLiteralOfType  and isLiteralNotOfType talk about IRI constants as the
>> second argument.
> 
> that's fine.
> 
>> I am happy with either keeping it like that or changing it, just wanted
>> to point out that there are two options.
>>
>>
>>> In fact, it would have been best if in BLD semantic structures the IRIs
>>> of datatypes are mapped to the corresponding datatypes, e.g., xsd:string
>>> is mapped to the XML schema string datatype.  One could then, in DTB,
>>> speak only about values and datatypes, which will be much more
>>> convenient and much more elegant.
>> I am not sure what you want to say here, can you explain/maybe
>> illustrate with an example?
> 
> I propose to extend the definition of semantic structure [1] by adding
> the following conditions to point 1 of the definition:
> - If a constant c \in Const is an IRI constant "d"^^rif:iri and d is a
> datatype identifier, i.e., d \in DTS, then I_C(d) is the datatype [2]
> identified by d.
> 
> Thinking again about this, we might get away with this change without
> redoing last call.  The only real implication it has is that equality
> statements of the form
> 
> xsd:integer=xsd:string
> 
> are currently not inconsistent, but with the proposed change they do
> become inconsistent.
> But we anyway don't want people to write this kind of statement; in
> fact, people should not use datatype identifiers outside of constants
> and isLiteralOfType/isLiteralNotOfType statements.
> 
> Best, Jos
> 
> [1] http://www.w3.org/2005/rules/wiki/BLD#Semantic_Structures
> 
>> Thanks,
>> Axel
>>
>>> We should not have moved BLD to last call before finalizing DTB :-(
>>> I now think we should probably redo BLD last call, after finalizing DTB.
>>>
>>>> Moreover, I think by dropping the specific guard predicates, we can get
>>>> rid of the definition of short names for symbol spaces as well.
>>> Yes.
>>>
>>>
>>> Best, Jos
>>>
>>>> Axel
>>>>
>>
> 


-- 
Dr. Axel Polleres
Digital Enterprise Research Institute, National University of Ireland, 
Galway
email: axel.polleres@deri.org  url: http://www.polleres.net/

Received on Monday, 2 February 2009 20:03:53 UTC