Lists Home |
Date Index |
> On a document level, it's easy to be draconian. Simply don't provide
> any xinclude:fallback elements. Then a missing resource will be
> fatal. The current draft doesn't provide any means to implement this
> at the parser level though.
Back to "the only way to get sane behavior is to rely on
non-standard extensions" evil. Chthulu is pleased!
> This is all new functionality in the
> recently released CR, so if you don't like this make a comment to the
> working group. (Personally, I think anything as major as this should
> require a return to working draft status. Stuff like this should not
> be added at CR.)
On the other hand, if the XInclude spec said that you needed
to substitute <xi:error code="..."/.>, rather than content found
in the document, it'd at least start to make technical sense as
a processor-neutral way to report errors ... even though it would
still be polluting the data content of the document. One could
imagine statements that documents with xi:error elements
have no canonical form.