Re: review rdf:text

> Herewith my review of the current version of rdf:text [1].
> 
> I have three substantive comments that I think should be implemented
> before going to last call:

Without studying them at them all that closely, these all seem like fine
changes (very nice work, Jos), but they also don't seem like the kind of
deep design changes that really need to happen before Last Call.  So if
the editors can do them this week, that seems fine, but if they have
trouble finding the time to get this done, I'd suggest letting these
change wait for the next round (hopefully CR).  Is that okay, Jos?

      -- Sandro

> 1- in the preliminaries: XML namespaces does not define a convention for
> the abbreviating URIs.  Rather, you are using the convention of RDF for
> abbreviating URIs.  In addition, the reference to XML namespace is
> misleading, since in XML names are pairs (namespace, local name),
> whereas in your document names are URIs.  Please refer to RDF when
> discussing the URI abbreviation mechanism.  Then, I believe the `.'
> after xpath-functions and rdf-text-functions should be replaced with a `#'.
> 2- section 3, definition of value space: I believe this is not a
> sufficient definition.  So, I propose to replace "contains" with
> "consists of"
> 3- definition of rtfn:compare: this function seems under-defined.  I
> believe that if $collation is given, and is not of type xs:string, the
> function should raise an error.  Then, it is undefined what happens if
> both comparands do not have a language tag.  In particular, it is not
> possible to determine whether their language tags are unequal, because
> they're not there.
> 
> After these comments are addressed, the document can go to last call, as
> far as I'm concerned.
> 
> 
> I have a few more editorial comments:
> 
> 4- Last paragraph of the introduction: the rdf:XMLLiteral datatype does
> not have a feature called xml:lang.
> 5- section 5.1.1: why write "of type rdf:text" twice?  The function
> simply returns these values, and clearly they are of type rdf:text.
> 6- in sections 5.1.2 and 5.1.3 I strongly suggest to replace "extract"
> with "returns", to make the definition of the function more homogeneous
> and because "extract" is not defined
> 7- the style specification of rtfn:compare is somewhat different from
> the specification of the other functions.  In rtfn:compare English is
> used to refer to different components of the input values, whereas in
> the other specifications symbols (s,l) are being used.  I would suggest
> to you symbols in section 5.2.1 as well.
> 8- in section 5.2.1, it is unclear what the mention "operators" are.  As
> far as I know, RDF, RIF, and OWL 2 do not define any of the mentioned
> operators.
> 
> 
> Jos
> 
> 
> [1]
> http://www.w3.org/2007/OWL/wiki/index.php?title=InternationalizedStringSpec&o
> ldid=21811
> -- 
> +43 1 58801 18470        debruijn@inf.unibz.it
> 
> Jos de Bruijn,        http://www.debruijn.net/
> ----------------------------------------------
> Many would be cowards if they had courage
> enough.
>   - Thomas Fuller

Received on Tuesday, 14 April 2009 11:56:27 UTC