- From: TAMURA Kent <kent@trl.ibm.co.jp>
- Date: Fri, 3 Dec 1999 13:50:39 +0900
- To: w3c-ietf-xmldsig@w3.org
I have comments for Transform(s) in the 19991116 draft. There are some types of information passed from/to Transform. A handling of `node-set' type, which is output of XPath, is not defined neatly in the draft. It seems that the draft says in `5.6.3 XPath Filtering' that a node-set always can be treated as an XML document. But it is not true because an XML document must have just one element, and a node-set may consist of more than one element, or attributes, or character data. We can not implement the first characteristics in 5.6.3 based on XML Infoset, that is, an XML processor does not provide information of an XML declaration, a character encoding, and existence of BOM in a document. I propose to define types of information as the following. ................................................................ There are three types: o Octet stream o Character stream o XML document (or single XML element) o Transforms IN: octet stream (for external resource), or XML document (an Object element, for embeded resource) OUT: octet stream o Encoding IN: octet sequence OUT: octet sequence o C14n IN: XML document OUT: octet sequence o XPath/XPointer IN: XML document OUT: XML document (if a node-set is single element), or character stream (if a node-set is single text node) Error (otherwise) o XSLT IN: XML document OUT: XML document (if <xsl:output method="XML"/>) character stream (if <xsl:output method="TEXT"/>) (Octet sequences generated by XSLT are implementation-dependent. For instance, XSLT document can not control order of attributes. So, a C14n is needed after XSLT Transfroms.) If a Transform expects an XML document as the input but the output of previous Transform is a octet sequence or a character sequence, the sequence is parsed as an XML document. If a Transform expects an octet sequence as the input but the output of previous Transform is a character sequence, the sequence is encoded with UTF-8. We must use a C14n or an Encoding as the last Transform. ................................................................ -- TAMURA Kent @ Tokyo Research Laboratory, IBM
Received on Thursday, 2 December 1999 23:51:21 UTC