RE: IRIs everywhere (including XML namespaces)

I remember the XML Schema WG discussing this issue.

It is a simple matter of layering.  The schema-loc is parsed into white-space-separated
tokens. These are then interpreted as anyURIs.  Therefore, they cannot contain
literal whitespace characters.

It is the fundamental nature of representing text that uses one set of delimiters
using text that uses another set, that when there is a clash you have to use
delimiters. It is no different, say than when an IRI includes the single quote
character and the IRI is inside an XML attribute value that has used the
single quote delimiter. That delimiter character is not available for use
in the IRI, and IRI delimiters must be used.

In XML we don't allow remapping of delimiters (as SGML does inside elements)
so that you could have a URI where whose interpretation rules took precendence
over the XML rules.  Instead, the XML parsing rules always have precedence.

Anyone who tries to embed arbitrary strings inside any format needs to be
aware of the delimiter conventions for that format.  


Cheers
Rick Jelliffe

Received on Wednesday, 30 October 2002 05:17:14 UTC