Re: CURIEs (was Re: Several minor problems in the grammar for the functional-style syntax)

On 23 Mar 2009, at 08:02, Ivan Herman wrote:

> Bijan Parsia wrote:
>>
>> 3) Whether to send a comment to the CURIE folks (which ought kick  
>> them
>> out of CR).
>>    I think we should saying:
>>        a) The spec must specifically allow for restricted sets of
>> characters in CURIEs and perhaps define some "natural" subsets, e.g.,
>> NCnames, SPARQL, etc.
>
> +1
>
>>        b) The spec must allow XML host formats to forbid xmlns style
>> CURIE prefix decs while having an alternative one.
>
> The text currently says:
>
> [[[
> When CURIES are used in an XML-based host language, and that host
> language supports XML Namespaces, prefix values MUST be able to be
> defined using the 'xmlns:' syntax specified in [XMLNAMES]. Such host
> languages MAY also provide additional prefix mapping definition  
> mechanisms.
>
> ...
>
> ]]]
>
> My understanding is that you do not want to have namespace  
> mechanisms in
> OWL/XML.

? We have namespaces, and have to. Every element in OWL/XML is in the  
OWL namespace. If someone wants to embed SVG or XHTML in annotations  
then they need namespaces as well. So we're stuck.

What I want to forbid is overloading the namespace for CURIE prefix  
declarations.

> If so, OWL/XML it falls under the negation of the first
> condition, doesn't it? Ie, if it does not use xmlns, then, well, the
> obligation of using xmlns for prefix definition is simply not  
> relevant.

See above.

>>        c) There should be a standard alternative declaration  
>> mechanism
>> in XML, e.g.,:
>>    <xml:curieAbbr prefix=".." expansion="..."/>
>
> The current text says 'may' define its own. I do not think that this  
> WG
> should take a position on whether the CURIE spec should define a
> standard alternative mechanism, because it is irrelevant for this WG  
> (in
> view of the previous point and if my assumption is correct on OWL/ 
> XML).

Your assumption is wrong. But even if it were correct, why should each  
XML format invent its own declaration mechanism?

> (Of course, Bijan Parsia, or the Un of M'ter, may decide to make such
> comments!:-)

Sure. But I think it's relevant to this working group. It's the  
difference between using:

	<owl:curiePrefixDecl...

and
	<xml:curiePrefixDecl..

I think that's a significant difference. Frankly, the CURIE spec is  
just punting (and badly) on a key aspect of CURIE use.

> B.t.w., just as an info: the RDFa task force is currently discussing
> adding a @prefix attribute with
>
> prefix="pr1:expansion1 pr2:expansion2 ..."
>
> that could be used on any element to define the prefixes for the dom
> tree under this element. Ie, no separate element, just an attribute  
> that
> could be used anywhere. The TF is still not sure whether the separator
> character will be ':' (like above) or '='.

That would work too, but is pretty much tuned for coping with text/ 
html as a host language. If you look at RELAX NG and Schematron, they  
use elements for their namespace-in-content prefix declarations. I  
prefer that for a number of reasons, including ease of manipulation  
(i.e., not having to parse a microsyntax).

Cheers,
Bijan.

Received on Monday, 23 March 2009 08:13:24 UTC