Re: [xml-dev] Whitespace character references and XSD normalization

[richard@inf.ed.ac.uk (Richard Tobin)]

In article <4506D7B1.1090405@globalgraphics.com> you write:

>My question is, if an attribute value includes a non-space whitespace 
>character reference, and the XSD for the attribute has the collapse 
>whitespace facet, does the character reference get replaced with a space 
>before any validation?  Or does the referenced non-space whitespace 
>character remain after normalization?

XML itself treats character references in attributes differently from
the characters they represent, so that (for example) &#10; in an
attribute will emerge from the XML parser as a linefeed, while a
literal linefeed will be replaced by a space.

But XSD works on the parsed document.  It has no idea whether a
character was a reference in the input or not.  And it normalizes
strings of a given type identically regardless of whether they are
attribute values or element content.  So it will see &#10; as a
linefeed character and collapse it.

-- Richard