- From: Carlo Sartiani <sartiani@di.unipi.it>
- Date: Wed, 31 May 2000 14:34:49 +0200
- To: www-xml-query-comments@w3.org
- Cc: sartiani@di.unipi.it
I read your working draft and I have two simple question: 1) A data model instance is defined as a forest of trees (maybe consisting of a single node). In Appendix B you defined two forms of ordering among forest nodes, highlighting that: "Given a single input document, a global, total order can be defined on nodes. In general, however, it may not be possible to define a global order, for example when a data model instance contains nodesfrom multiple input documents." In Quilt definition paper (Sigmod/Pods/WebDB2000) authors, instead, explicitly stated that on every data model instance a total order relation is defined: "Each instance of the XML Query data model (regardless of whether it is a complete document, a fragment of a document, or a list of documents) is a forest that includes a total ordering, called "document order," among all its nodes." Are Quilt authors wrong or, instead, do you think to enrich XML Query Data Model with a mechanism for defining a total order relation among nodes coming from multiple input documents? 2)XML Query Data Model represents an XML document as a node-labelled tree, while traditional semistructured database data models exploit edge-labelled graphs (e.g., XML-QL). What are the main reasons behind your choice and what are the main advantages you think to obtain by using a tree representation instead of a graph representation? -- Carlo Sartiani Ph.D. Student Computer Science Department University of Pisa "Here's to the crazy ones. The misfits. The rebels. The troublemakers. The round pegs in the square holes. The ones who see things differently. The ones that change the world!"
Received on Wednesday, 31 May 2000 08:37:56 UTC