Comment on XLink WD: Traversal

Comment from W3C member Academia Sinica:
- Rick Jelliffe

The current draft fails to address traversal issues adequately.
Consequently it requires some other, unprovided, layers of software or
specifications in order to be useable. Furthermore, it is to be expected
XML data may be published in a different schema to the one required by
an application.  So it fails to meet the first of its design principles:
"1.1 XLink Shall Be Straightforwardly Usable Over the Internet"

To remedy this, I suggest the following:
    1) an attribute xlink:traverseFunction be added to locator elements,
with the declaration
            xlink:traverseFunction %URL;  #IMPLIED
    2)  the default behaviour for traversing a link is to use the
xlink:traverseFunction URL to as the base URL, the xlink:role value as a
CGI query name, and the href value as the parameter.  Traversal is thus
defined not as a GUI behaviour, but as a function on the XPointer.

For example, default traversing of:
    <xlink:simple
        href="http://www.sinica.edu.tw/sinofrog.xml"
        role="frogList"
        traverseFunction="http://www.geotempo.com/servlet/findFrogs"
    >Ribbit</xlink:simple>

will generate the following CGI:


http://www.geotempo.com/servlet/findFrogs?frogList=http://www.sinica.edu.tw/sinofrog.xml

This simple addition would result in almost no additional complexity,
maintain the separation between traversal and pointing, integrate into
the WWW better, and provide very useful functionality.  Most
importantly, it will make XLinks immediately useful without having to
wait for some other layer to be specified.  Locators with values that
are URNs would become useful, also.

Received on Wednesday, 29 December 1999 14:48:57 UTC