========================================================================
Development issue list for the Informa library project
$Id: TODO.txt,v 1.39 2006/01/03 00:30:36 niko_schmuck Exp $
========================================================================


WHAT REMAINS FOR NEXT RELEASES:


o integrate patches from Brian King
  - compare each implementation file, for example:
  - check mapping of Item <-> ItemSource Relationsship
  - merge also ChannelBuilder
  
  Finally check:
  * consistent equals() and hashCode() methods
  * all mapping of link attributes as strings
  

o integrate exporter refactoring from Peter Darton
  - patches/peter

o try to use hibernate-unittest.properties for TestHibernatePersistenceManager



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



o switch from jdom to dom4j for XML parsing

o integrate FeedManager with OPML parsing

o proceed with ChannelBuilder restructurement:

  builder = "fqcn".newInstance();
  builder.init();
  ...
  builder.close();

o evaluate further convergence between ChannelRegistry and ChannelBuilderIF

o integrate hypersonic SQL as test (default?) data storage for the 
  hibernate backend

o split user metadata out of core interface usages, that is:
  ChannelIF
  - getSubscription
  - setSubscription
  ItemIF
  - getMetadata
  - setMetadata

o ChannelGroupIF add composite pattern, so that groups may have
  zero or more ChannelGroupIF objects as children

o check that new Channel/Item categories assignment works with castor
  JDO write some test cases for this, check also im-/exporters

o fix long transaction problem
  implement org.exolab.castor.jdo.TimeStampable interface:
  - long jdoGetTimeStamp()
  - void jdoSetTimeStamp(long timeStamp)
  to allow Castor dirty-checking

o improve minigui.py for easier restarting from existant
  channel structure / data

o how can minigui use castor JDO implementation ?

o think more about extracting IdGenerator from concrete
  object instantiation, put it into ChannelBuilder for example

o what to do about HTML markup inside item title/description fields
  - generally
  - when indexing (extracting first the content (?))
  - rigurous strip-off markup ??

o if dc:date available than make use of it, instead of using the
  default found behaviour.  ---> for better support of <dc:date> (of
  indivdual item) it could make sense to move found date into
  ItemSubscription and have a (second) date field (just "date") instead

o test/jdo/marshal.py: 
  - investigate how to best marshal a group of channels, since query: 
    SELECT g FROM de.nava.informa.impl.jdo.ChannelGroup g
    results in <group id="1"/><group id="2"/>...
  - when marshalling channel-by-channel is it really necessary to
    have one enclosing root element, how would this be produced?

o support RSS 0.92 fully (fix then fake behaviour of FormatDetector)

o add some more test cases
  (especially test castor JDO un-/marshalling)

o improve how channels are retrieved (HttpClient library from jakarta)
  for better socket timeouts and client behaviour (for example
  setting request header to something different than "Java/1.4.1_02")
  http://jakarta.apache.org/commons/httpclient/ 


--- documentation / web-site ---

o describe setting-up and usage of hibernate backend

o add FAQ about:
  - how to export to a stream (wrapping a stream into a writer object)
  - sorting items

o clean up docs directory 
  - remove NOTES.txt ?
  - integrate resources.txt into rss-feeds.txt

o finish class diagram of the core interfaces for easier
  explanation (use Poseidon UML to generate diagrams?)

o castor JDO mapping should be documented and made available also on
  the web-site

o short example using castor JDO would be nice

o short example about how to incoporate full-text search with Lucene
  would be also nice


------------------------------------------------------------------------

ISSUES TO BE ADDRESSED BY FOLLOWING RELEASES:

o write ItemCleaner, which would remove all items which are older
  than a certain date

o add convenience method to ActionRegistry for updating one channel
  without repetition (also updating all channels at once)

o make the observing of multiple channels more elegant, instead of
  adding an observer to each channel ???

o improve fault tolerance, do not throw RuntimeException if not really
  necessary (for example, in the case a stream can not be found, just log
  warn message and try next time)

o more test cases:
  - add parser test for partly empty TextInput and Image elements
  - equals test for Channel and Item
  - may be some JDO load tests (very interesting to watch it scaling)

o may be have a utils.XMLPersistenceManager (using Castor XML) and
  a dedicated mapping.xml (writeRegistry, readRegistry)

o common super class for similar ChannelParsers (using JDOM) ???
  (not sure this makes really sense)

o check if subject (member of Item) should be rather a collection
  of SubjectIF objects instead of string (figure out more about
  how and where subject is used) to improve support for RSS 1.0

o more thinking about FeedIF: similar members needed like in ChannelIF?
  format, subscription (?)

o OCS integration: writer parser and generate channel structure

o OPML integration (nested outlines containing links) used by UserLand


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

Please use CHANGELOG.txt to note the changes you made.

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