SAX2: summary of Namespace-support arguments

Rick Jelliffe ricko at
Sun Dec 19 07:28:10 GMT 1999

 From: James Clark <jjc at>

 >I do indeed want that, and in the past I've argued against *requiring*
>processors to provide information about the prefix used. I've become a
>lot less negative about this prefix information recently and I think
>it's better for an API to provide it

Isn't it a requirement for XSL?  In XSL, if two elements have the name
namespace but different prefixes, they are processed differently (at
least, for the XSL namespace itself).

This is essential for XSL scripts that generate XSL.  This cannot be
done if either the prefix is lost or unless some other arrangement
(e.g. adding some attribute giving an instance number) is used.

Also, prefixes form part of the markup.  They should have documentation
meaning.   I think it is bad if the "rdf": prefix were to be replace by
"xyz" or, worse, by some prefix generated at random like "dom".   So,
like document-entity encoding, prefixes do not form part of the
information set of the document but they do form part of the pragmatic
information set required for round-tripping.

Different instances of the same document are required for programming
languages that use XML syntax and which allow namespaced elements
as data.  Prefix-preservation is required for this. Prefix-preservation
is also required for round-tripping.

Rick Jelliffe

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at
Archived as: and on CD-ROM/ISBN 981-02-3594-1
To unsubscribe, mailto:majordomo at the following message;
unsubscribe xml-dev
To subscribe to the digests, mailto:majordomo at the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at

More information about the Xml-dev mailing list