TODO.txt

* Make sure that if there is a default value for a class that it is used
  when none is provided (when reading).
   
* Update all project READMEs.

* When serializing, check if source property declaration is an interface
  and whether the written type is declared as a subtype already. If not,
  write "class" with fully qualified name.  Make sure FQN are supported
  on reading as well.

* Do a swipe of all class/method names to apply case-sensitivity in 
  a consistent way when using acronyms (e.g., XmlBlah vs XMLBlah).

* Redo all metadata constants to be more consistent in naming and syntax,
  plus remove references to "collector.".  Make them all prefixed the same,
  including Importer (e.g., "doc.*")

Configuration:
--------------- 
* Maybe, have a Configurable interface with set/getConfiguration method
  which will get autowired with the matching class.
  or look at autowiring framework as a global addition.
  Or, have annotation @config that will take the type of the property
  it decorates, or accepts an optional configuration object as a value.
  Dagger 2 seems best: well-maintained, lightweight and compile-time

* Consider replacing XSD with OpenAPI JSON Schema.  That way, using the crawler
  as a service could automatically build a swagger api and clients, in 
  addition that users can use a bunch of existing json-schema-based editors.
  We could also integrate https://github.com/rjsf-team/react-jsonschema-form
  in our SaaS app.
  MUI5 example: https://rjsf-team.github.io/react-jsonschema-form/
  
* Have a Server mode, that would allow crawling of passed URL mixed 
  with config snippets, based on JSON schema.  

  

* Remove all "restrictTo" across the board in favor of XML flow (if/else, etc).
  At a minimum, if we keep it, wrap in "restrictions" to improve null handling.
  (e.g., blanking a default list)

* Add this https://www.mojohaus.org/versions/versions-maven-plugin/examples/display-plugin-updates.html

* Remove all @since prior V4 since the package change
  makes all classes "new" anyway?
  
* Remove all @author tags in favor of Git commit author?
  
* V3 to V4 config converter? or add a flag to support V3
  config as argument?
  
- Consider renaming (and deprecating) classes/methods with 
  acronyms/abbreviation to use camel-case.  
  E.g. XmlSomething as opposed to XMLSomething.
  That includes in Commons Lang
  
* For javadoc custom taglets, add a placeholder equivalent to <THIS_CLASS>
  that is replaced by the fully qualified name of the current class. 

* Maybe have a data store just for general session/crawl state 
  persistence, not just documents. This would allow us to store things
  such as :
    * Was queue initialization complete (useful to know when resuming).
    * What else? 

* On online documentation, add a flag that indicates whether a provider
  of something is "nullable" as a way to disable it.  Else, an exception
  is thrown (nullpointer or whatever).  
  
* Add extra validation for invalid config state.  For instance, 
  specifying a sitemap start URL and disabling sitemap support should
  prevent the crawler from doing anything.