Tags allowed in xdocs style documents:

http://jakarta.apache.org/site/jakarta-site-tags.html


---

from build.xml, the convenient way to include tests in subdirs:

      <batchtest todir="${test.data}">
        <fileset dir="${test.classes}">
          <include name="**/Test*.class"/>
        </fileset>
      </batchtest>

=================================================================


NOTE: This is out-dated please see the TODO for further details.


README ----------------------------------------------------------
A very rough sketch for what the RSS Library could look like
-----------------------------------------------------------------

[Note: This is very far from being complete, just my very first ideas
thrashed out of my brain, especially the names of the classes and 
package names are highly in-flux.]

- how parsing could happen with the proposed interfaces:

  the application would held a collection of FeedParser objects (at the
  beginning may only a parser for OCS) and URLs to (OCS) news feeds.
  
  Some pseudo code for retrieving a list of available channels:
  
  // present the channels to the user and let him/her decide
  // wether to subscribe to a channel or not
  channels = FeedParserIF.parse(ocs_url)

  The user could of course add manually URLs to channels he wants to
  subscribe to. To get all the items of the subscribed channels
  please conside the following (again this is pseudo code):

  // channels is a collection of ChannelIF objects
  for each channel in subscribed_channels:
    channel_format = channel.getFormat()
    parser = channel_parser_collection.getParser(channel_format)
    items = parser.parse(channel.getLocation())


- possible implementations of ChannelParserIF:
  - de.nava.informa.parsers.RSS_0_91_Parser
  - de.nava.informa.parsers.RSS_1_0_Parser


- possible implementations of FeedParserIF:
  - de.nava.informa.parsers.OCSParser


- the core interfaces belonging to the "news object model"
  could be implemented in-memory and in a SQL manner
  (may be going via a O/R mapping tool)?
  
  
- utility classes would allow to write out (serialize)
  RSS Channels (also in different formats), may be we need
  a ChannelWriterIF interface ?


- questions:

  - rename ChannelIF#contentType to format ?

  - support single-user, or allow multi-user to access the environment (do
    we want that this library is only used by Stand-alone applications
    or also be used by multi-user environments like web-applications) ?

  - state handling inside or outside the interfaces (may be best to separate
    them) ?

  - general: should the IDs be integers or String objects ?


- missing interfaces:

  - MonitorIF: for handling poll threads (postponed)
  
  
- implementation notes:

  - Item: hashCode important for comparing if item already exists
          (Name/Title and Link/Location), do not take parent channel
	        into account (since this may lack when item is not assigned
	        because it has to be checked wether it is also already existant)


Feedback is very welcome :)


Niko Schmuck (niko@nava.de)
Created: 26.04.2002
$Id: developer-notes.txt,v 1.3 2003/09/26 14:39:40 niko_schmuck Exp $