Re: xml syntax issues

Sandro Hawke wrote:
> In case we get to the XML Syntax discussion in the meeting, I've put
> together a list of "coin-flip" [1] decisions that have to be made here (8
> of them as of this writing, although there are some subsidiary
> decisions).
> 
>    http://www.w3.org/2005/rules/wg/wiki/Arch/XML_Syntax
> 
> My expectation is for this page to evolve in place, to have resolved
> details along with any remaining issues.
> 
> Feel free to add attributed comments to the page (you'll see some from
> me -- they start with "SandroHawke") or reply to this e-mail.

Thanks, I've annotated the page with my (mild) preferences.

There was one other issue I nearly added but then thought perhaps it 
doesn't quality as a coin flip (though it's not exactly a deep issue 
either) - whether we allow curie syntax for IRIs.

Use of full IRIs (other than ones relative to the xml:base) does make 
the syntax particularly unreadable (people do need to sometimes be able 
to look at the serialization).

   <Dog iri="http://hawke.org/2005/Taiko">

CURIEs (qname-like syntax for defining a URI/IRI) help a bit:

   <Dog iri="[ns:taiko]">
or
   <Dog curie="ns:taiko">

On the other hand people can always use XML entity references to 
abbreviate IRIs, not fantastically readable either but equally short:

   <Dog iri="&ns;taiko">

I think I'm inclined towards the simplest case of not having CURIEs in 
the XML. However, I do think we want CURIEs in the linear ("human 
readable") syntax so it's reasonable to at least consider whether they 
should also go in the XML.

Dave
-- 
Hewlett-Packard Limited
Registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England

Received on Tuesday, 15 May 2007 08:30:02 UTC