Tim Bray wrote:

> At 12:14 PM 11/18/99 -0700, Tom Palmer wrote:
> >Still, what if someone wanted to use one organization's parser
> >and another's DOM implementation?  These should be kept
> >independent if at all possible.
>
> Errr, how is that possible?  The way things normally work is, parser
> reads XML resource, loads it into data structure, offers DOM API to it.
> The parser and DOM implementation are necessarily somewhat in bed, unless
> we want to require DOM implementations to use SAX to build it.  Do we?  -T.

There are many, many use-cases for DOM, indepenendent of any parser.
For example, IBM's XML Productivity Kit for Java (available on
http://www.alphaworks.ibm.com) is a set of Java Beans manipulates XML, and
passes the result from one bean to the next.

The DOM beans receive a DOMEvent (a wrapped Document reference) usually do some
transformation or filter and send a DOMEvent as output. The LotusXSL processor
is included which takes 2 DOMEvents (or Document references) one for the input,
one for the rules, and outputs the result as a Document reference.

So the parser can be superfluous. The DOM tree is simply an API, and the source
could easily be a database, or non-XML format. So you could read in some
deliminted format, programmatically create a DOM tree using a generic DOM,
manipulate it and then invoke some parser-independent DOM writer, printer to
create XML or whatever.

-Ralf


Reply via email to