Revision history for Perl extension Lingua::LinkParser.
0.01 Thu Jan 13 09:54:28 2000
- original version.
0.05 Tue Feb 8 13:00:13 2000
- public beta
1.0 Wed Aug 16 14:35:44 2000
- Rewritten for the new Link Grammar 4.0 API
- Got LinkParser.xs to work with the char** functions, so
linkage_get_words and get_domain_names now work.
- Makefile.PL changes LinkParser.pm to find data directory
- new() expects a hash defined the the pod documentation, and uses the
the global $DATA_DIR to as a default path
- Added Lingua::LinkParser::Definitions to enable easy retrieval of
link summaries and definitions from within the program. This is a
totally non-orthoganal method of distributing this info, but I find
it handy.
- Added linkage_get_word to ::Linkage, thanks Richard A. Wells.
- Added linkage_get_words, also thanks to Richard.
- Added access to the domain functions with num_domains and domain_names
to the Lingua::LinkParser::Sublinkage class.
- Added function 'violation_name' to Lingua::LinkParser::Linkage class to
return the label for a post-processing linkage violation.
- Added the new 'print_constituent_tree()' from the 4.0 API to display
constituent tree structures.
- Added tests to 'test.pl' for 'get_words()', 'num_domains()', and
'domain_names()'
- Added scripts to demonstrate use of 'bigstruct()' and 'define()'
- Added script to demonstrate 'print_constituent_tree()'
- Included the link guide in directory 'guide', since it's not in the
distribution.
- Overloaded the stringify operators ("") for classes ::Sentence,
::Linkage, and ::Sublinkage to display a diagram for each.
- Changed 'get_postscript' to use new 'mode' parameter
- Added the link functions that had previously only been available to
sublinkage objects to linkage objects.
1.01 Wed Sep 2 12:35:01 2000
- Built and tested using Link Grammar 4.1
- Fixed bug that redefined 'get_word' and 'num_word' functions in
Linkage.pm.
- Moved Dictionary 'type' to its own namespace,
Lingua::LinkParser::Dictionary.
- Removed DESTROY methods from ::Sublinkage and ::Linkage; the API
4.0 was crapping a log when invoked.
- Added a class 'Lingua::LinkParser::Linkage::Word' to allow word
object creation and quick access to information for that word.
- Lots to fixes to ::Sublinkage to get index numbers right
- Lots of examples added to 'scripts/', including a grammar regex
- Added alternate methods for overloading ::Linkage and
::Sublinkage, "new_as_string()", to experiment with grammar regex
- Changed overloaded behaviour for ("") constants for ::Sentence to
print out the original sentence text.
1.02 Tue Sep 5 23:42:58 2000
- Added XSUBs for 'max_sentence_length' opts.
- Fixed Makefile.PL to find data file paths (link parser 4.1 still has
dictionary files named "4.0")
- Changed '::Linkage::opts()' to take a hash of keys/values, or returns
a value of only one argument is passed.
- Modified the script 'scripts/parse.pl' to parse null linkages
automatically, and demonstrate more elements of the parser options.
- Changed the default behaviour of the overloaded ("") for ::Linkage and
::Sublinkage to use 'new_as_string'.
1.03 Mon Nov 6 11:46:28 2000
Thanks to Ronald J Kimball for recommending the following changes:
- Made changes to Linkage.pm and Sublinkage.pm to avoid uninitialized
value warnings.
- Fixed grammarregex.pl to use a negated look-ahead assertion.
TODO:
Change constructors to verify arguments.