Design

• Several XSLT stylesheets (for formatting) managed via several xsl-stylesheet processing instructions
• two stylesheets applied to the document: one for querying the data, one for formatting
• the XPath generator GUI makes use of object-orientated Javascript programming (Reference book: Javascript - the definite guide, by David Flanagan - O'Reilly publisher); it has an object model of an XPath string
• use of the terminal part of the URL to obtain the URL of the XML data, e.g..:

http://wwbota.free.fr/Generic/XMLClient.htm?URL=http://www.mycom.com/mydata.xml
This is in the reverse way to the normal use (i. e. from URL to browser, instead of from browser to server via URL).
• the analysis of the document ("implicit schema") makes use of a generic DOM iterator following the Visitor design pattern (function applyRecursive); since in Javascript a function is also an object it allows to gather any information in the visiting function itself.

Possible enhancements

• enhance ergonomy
• better representation of the schema (maybe with embeded HTML lists)
• Add buttons : "Remove XPath Step" "Remove XPath constraint";
• make a pulldown menu for n possible XSLT stylesheets associated to the XML data;
• update the generated XPath string at each user input, or very often via a setTimeout() call

• suppress IE non-portable Javascript items: document.all , for= , ...
• It should be possible to adapt this page for other browsers, either with the Xalan applet, or with Mozilla's  new  XSLT functionalities;
• Also desirable would be the generation of a skeleton DOM document to record the structure of the instance document, instead of the simple associative array with XPath strings that is currently used.