"Proper" way to use XML (not library specific)

[Will Sappington <ws...@ndma.us>]

Hello again Dave, and all,

Apologies in advance if this is too long, but this is about as briefly
as I can describe the issue clearly.

The company I work for is standardizing on XML as the means for
representing various types of data, our configuration files being one of
them.  We currently use what we call an application profile, just a
hierarchical structure of name/value pairs, organized in an
application/section/item hierarchy.  The functional interface to the
profile is 1) open(), which opens the file and loads it into memory, 2)
execute one or more getItem()'s to pull out configuration items
specified by application-name/section-name/item-name, and 3) close() the
profile.

I'm new to XML, but based on a recommendation and my own analysis after
the fact, XPath seemed a reasonable way to replicate this functionality
in XML because it allows you to directly access specific elements.  I've
had some trouble implementing this, mostly due to a lack of
understanding of the library that was chosen (Xalan/Xerces).  In
response to my troubles, my manager is saying that the problem is with
how I'm trying to use XML, not with the library.  He says XPath is not
appropriate for this, that I should "marshal" the entire XML file into a
different form, toss the DOM, and operate on the transformed data.  I'll
put his email to me, edited for brevity, below.  But my question is, if
XPath isn't appropriate for this, then what is it appropriate for?  From
the perspective of a configuration utility, it certainly seems to be
reasonable, if not obvious for the interface to be such that the user
can read specific configuration items from the file in whatever order is
desired, and XPath appears to be designed to do specifically that with
XML data.  So why would it not be an appropriate tool to use for
migrating our existing name/value pairs (.ini files) to XML?

Many thanks in advance for any thoughts you might care to share.

-will 

(edited (snipped) comments mentioned above)

The problem is with the way you want to use the library.

You have been working way too long with things like configuration files
where you read out the parameters as you need them.

Here is what you need to do:

Create a class to hold the information contained in the XML. 
Next load the profile into the class. 
Extract the information from the XML into the class. 
Now dispose the DOM tree, and use the class for all references to the
parameters you need. 

What this is called is marshaling the data. If you want, you could
download an evaluation copy of an XML tool and it would generate the
code to do all that for you.

This is the way most people use XML. You need to learn to do it this
way.


The system reads the XML file one and only one time. The code loads the
data from the DOM tree then gets rid of the DOM tree.