Status update.

[James Bognar <ja...@salesforce.com>]

Hi all,

I've been working on tightening up the code by trying to convert the code
to use final fields where possible.  It seems to be having a small but
positive impact on performance.


There are two changes I'd like to make to the code....on minor and one
major....


First, I'd like to replace @BeanProperty(beanUri=true) with
@Rdf(beanUri=true).  Only RDF has the concept of bean URIs, so I'd like to
move it out of the common code.


Second, I want to introduce the concept of class lexicons.  A class lexicon
is a simple mapping of names to classes to help identify class types when
they cannot be inferred through reflection (e.g. when parsing into an
Object or superclass/interface field).  The names would also be used as
element names in XML.

Class names would be defined by the following annotations...
@Bean(name="foo") - For beans.
@Pojo(name="foo") - For other POJOs.

Lexicons would be defined by the following annotations....
@Bean(lexicon={list-of-classes})
@Pojo(lexicon={list-of-classes})
@BeanProperty(lexicon={list-of-classes})
...and also serializer/parser properties...
SERIALIZER_lexicon, PARSER_lexiconClasses

For example....

@Bean(name="foo")
public class Foo {
   @BeanProperty(lexicon={Bar.class,Baz.class})
   public Object[] x = new Object[]{new Bar(), new Baz()};
}

@Bean(name="bar")
public class Bar {}

@Bean(name="baz")
public class Baz {}

When serialized as XML, you would see...
<foo>
   <x>
      <bar/>
      <baz/>
   </x>
</foo>

When serialized as JSON, 'n' attributes would be added when needed to infer
the type during parsing...
{
   x: [
      {n:'bar'},
      {n:'baz'}
   ]
}


This change should allow me to eliminate the following features that I was
never happy about:
1) The "_class" attribute.  Will be replaced with "n" attribute.
2) The @Bean(subTypeAttr) and @Bean(subTypeAttrs) annotations.


-- 
James Bognar