How to do marray Torque and JaxMe?

[Jochen Wiedmann <jo...@ispsoft.de>]

Hi,

I would like to ask for advice on how to do the following.
First some words of introduction.

JaxMe 1 is a Java/XML binding framework. It takes an XML Schema
as input and creates sources that allow to marshal and/or unmarshal
the generated Java objects into XML documents and vice versa.
It also has a persistence layer: The generator allows to read
the schema of an SQL table and add the columns to the schema
at a designated place. The generated sources allow to read data
from the table into the generated Java objects and vice versa.
A concept which is well known to Torque users. See
http://jaxme.sf.net/ for details on JaxMe 1.

I am currently working on JaxMe 2. The main difference is
that it follows the JAXB specification (see http://java.sun.com/)
and it supports a much larger subset of XML Schema. It doesn't
yet have a persistence layer. I am currently considering how to
do this. IMO an excellent approach would be to use Torque as a
base. See
http://cvs.sf.net/cgi-bin/viewcvs.cgi/*checkout*/jaxme/JaxMe2/docs/Design.html
for details on JaxMe 2.

However, I have never looked into the Torque sources. It could
probably help me a lot if some experienced Torque developer would
tell me a general guideline on how to do this:

  - Create an Ant task very similar to sql2xml. This Ant task
    would take an XML Schema as input. At some point of the
    schema it would find the words "insert the columns of table
    foo here", which is exactly what the task would do. The
    main difference to sql2xml would be that no schema would
    be emitted, but JaxMe's internal representation of an XML
    Schema.
  - Create an Ant task very similar to "sql" that takes the
    above internal representation as input and generates
    similar sources. The main difference would be, that the
    target objects would be "JaxMe objects". JaxMe objects are,
    in general, Java Beans. However, they might have additional
    child elements or attributes, which are not covered by the
    columns. They also might have mixed content or whatever
    appropriate for XML.


Regards,

Jochen