The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
0.15  8.14.2004

- Changed hard-coded 'D' and 'T' prefixing to use _nodeify subroutine
- Added connected_components method 
- Began adding support for file-based hashes using MLDBM, enabling very
	large collections.
- Minor bugfixes, now tests for D: and T: prefixes explicitly,
	instead of assuming that those are the only node types.
	
0.14  7.13.2004

- Fixed bug in orphan removal on node delete operation
- Updated 'store' method to use network order
- Fixed bug where spurious 'D:' and 'T:' in doc titles misled 
  term/doc list methods
- Added find_by_title method and tests
  
0.13  3.14.2004

- Changed interface on 'delete' to support term node deletion
- Updated invariant.t and add_subtract.t with interface change

0.12  3.13.2004

- Added intersection method
- Added merge method

0.11  3.11.2004

- Added add_file method at suggestion of Aaron Cope
- Fixed bug in term_list method that would always return full list

0.10  3.03.2004

- Made adding content-free document a fatal error
- Added simple_search method
- Added load_from_dir constructor
- Added dependency on File::Find
- Normalize all terms to lower case

0.09  2.23.2004

- It's now possible to add, delete, and update individual docs
- Some regrettable but correct interface changes
- Fixed serious bug where weighting was not being recalculated
- Graph can now be rebalanced by hand
- Adjusted graph to store lcounts as well as weights
- Made global weighting optional
- Broke C implementation :-(
- Added new test suites

0.08 ( not released )

- Reduced size of Node struct in XS code (removed type field)
- XS and Perl code now preallocates space when reading from TDM file
- Added max_depth attribute, to limit search depth

0.07

- Altered energy distribution scheme to more effective log(degree)+1 
	(thanks to Martin Spernau for testing this)

0.06  7.23.2003

- Merged in XS implementation from Schuyler Erle
- Fixed global weighting bug where counts and weights were being computed in
	the same loop
- Updated test cases to reflect bugfixes
- Added XS/pure Perl switch in constructor
- Updated test cases to test both XS and pure Perl versions

0.05  7.17.2003

- Module now inherits from Storable.pm
- Added "find similar" method for document similarity search
- Added "mixed_search" method for term + document search;
- Edges are now normalized around each document
- Renamed TDM load method to "load_from_tdm"
- Singletons now inhabit the graph, instead of living in a lookup table
- Special check prevents bounce-back from singletons
- Many more test cases
- Two levels of debugging output
- Added credits for John Cuadrado and Scott Preece	

0.04  7.14.2003

- Applied patches from Leon Brocard
	 - actual test cases
	 - fixed typo that broke module
	 - cleanup of debugging flag and output
	 - added dependency on Test::More

0.03  7.11.2003 

- Added check to make sure no edges can have greater than unit weight

0.02  7.11.2003

- Added interface for just adding document titles + word lists, and having the sucker
  build the search engine for you.   Earlier TDM format universally reviled by 
  Perl community...
- Added some POD and minor code patches from Ken Williams

0.01 6.01.2003

- And Maciej said, "Let there be Search::ContextGraph"!