RE: [xsl] attribute nodes

["Christopher R. Maden" <crism at maden dot org>]

At 22:29 4-05-2001, Michael Fitzgerald wrote:
>Thanks all for answers. In the traditional tree data model in CS, a good
>forty years old at least, "if p is the parent of node c, we also say that c
>is a child of p." [1] XSLT/XPath in a sense breaks this traditional
>relationship wrt attributes, perhaps only as a convenience for tree
>traversal. I don't know. I'm asking.  -Mike

And Michael Kay answered, accurately, to my recollection.  What additional 
information are you seeking?

You'll note that the Aho & Ullman definition appears almost verbatim in the 
XML 1.0 Recommendation.  It applies only to elements - a child element has 
a parent element, and a parent element may have child elements.  XPath 
adopted that model, and extended the terminology where it was logical: for 
comments, processing instructions, and text nodes, which are ordered within 
the three.  Attributes are not children; I've never seen a discussion of 
XML that describes them as such, and they don't fit with the observed user 
expectations of children.  They are illustrated as decorations on a tree or 
tags on boxes, not as descendants or nested boxes.

But this presents a problem.  We have the attribute axis for referencing 
attribute nodes, distinct from the child axis.  But an attribute has an 
"owner" that it is interesting to discuss; what do we call it?  There was 
some considerable debate about breaking the symmetry of child and parent, 
but for simplicity and least surprise (except among hard-core CS nerds), 
the final decision seems to have worked well.

-Chris
-- 
Christopher R. Maden, XML Consultant
DTDs/schemas - conversion - ebooks - publishing - Web - B2B - training
<URL: http://crism.maden.org/consulting/ >
PGP Fingerprint: BBA6 4085 DED0 E176 D6D4  5DFC AC52 F825 AFEC 58DA


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list