ChangeLog for NCL 2.1

================================================================================
changes in svn versus previous tagged version
================================================================================
================================================================================
2.1.18 (relative to 2.1.17) SVN revision 641
================================================================================
	- As of March 09, 2012, NCL is available under a Simplified BSD license (see
BSDLicense.txt) in addition to the GPL license.

	- improved use of std:: decorator (to improve portability).

	- bug fix in data pattern compression routine when a charset is passed in - this only affects client code that this using NCL to compress data matrices.  
	
	- bugfixes for reading newick trees (including a still-hacky fix for dealing with numeric labels that cannot be interpreted as indices)

================================================================================
2.1.17 (relative to 2.1.16)
================================================================================
New features:
	- Warning suppression (mainly on integer conversion).  Thanks to John 
		Huelsenbeck!
	- config.guess updated using  automake v1.11.1 in response to report of user
		error by DZ.
	

================================================================================
release that will be 2.1.16 (relative to 2.1.15)
================================================================================
New features:
	- TaxaAssociation block now supported and treated as a "public" block
	
	
================================================================================
2.1.15 (relative to 2.1.14)   Revision: 617
================================================================================
bug fix: 
	- NxsReader returns the most-recently read block read by default (as the
		error message has always indicated) rather than the first block. This
		only affects blocks without link commands and ambiguous 
		references to blocks.
	- typo fixed in error message for invalid fasta file reporting
	
New features enhancements:
	- configure argument --with-reserveid added, so that the variable 
		name id will not be used. This avoids conflict with the objective C 
		keyword "id". thanks to John Huelsenbeck for the bug report
	- non-recursive NxsSimpleNode::AddSelfAndDesToPreorder (for huge trees)
	- refactored NxsSimpleNode to add RerootAtNode function

================================================================================
2.1.14 (relative to 2.1.13)   Revision: 611
================================================================================

Changes breaking backward compatibility with v2.1.13:
    - None
    
    
API changes:
    - Moving several functions if optional tree processing from private to public
    
    - Changed some of the less-widely used example programs to noinst_PROGRAMS 
        rather than bin_PROGRAMS so that we are not polluting the /usr/local/bin dir 
         of every user

New features:

    - NxsTreesBlock::SetTreatAsRootedByDefault(false) can now be called so that 
    	trees read will be treated as unrooted by default (this setting will be
	    ignored if the tree is explicitly labeled as rooted via [&R] comment or
	    unrooted via UTREE command or [&U] comment).
    - NxsToken::Tokenize static function added (takes a string and returns a vector
        of tokens.
    - SetCoerceUnderscoresToSpaces added to MultiFormatReader
    - GetNumLabelsCurrentlyStored added to NxsLabelToIndicesMapper to help with parsing
        numeric trees
    - Added optional treatIntegerLabelsAsNumbers setting to NxsTreesBlock for reading
        numeric trees
    - Added some functions for compressing redundant columns of a data matrix (see
        the new example
    - More command line flags supported by NCLConvertor, NEXUSnormalizer and gap 
        code

Bug Fixes:
    - hasPolytomies field of NxsFullTreeDescription is now set correctly. 
    - parsing of weight set of assumptions block
    - configure now fails (as it should) if CXX is not working 




================================================================================
2.1.13 (relative to 2.1.12)   Revision: 575
================================================================================

minor API changes:
	- nxscxxdiscretematrix ctor now takes an optional set of indices to include.
	- It was convenient to make some of the error reporting functions more 
		generic, so the NxsException class was given a addPositionInfo method,
		and some of the characters block code was slightly refactored (to not
		demand a NxsToken object for some calls).
	- specified the facet values for the DataTypesEnum so that it will be easier 	
		for client code written in non-C++ languages to interpret GetDataType()

bug fixes:
	- fixed strange pointer equality check that had used subtraction (and was
		causing problems on new windows g++ - v4.2 64bit).
	- explicitly casting return of size() to an unsigned to avoid g++ issues
		with the NxsString << operators.
	- You now must compile with HAVE_NCL_NXSSTRING_ENDL defined if you want the
		funky NxsString&operator<<(NxsString &(*funcPtr)(NxsString&)) operator.
	- include of <iterator> for vis studio 2010
	- Changes to allow compilation in Visual Studio.
new features:
	- Added support for the .fin files that are posted by Goloboff et al in
		their  Cladistics 25 (2009) 211230 paper.
	- better error message for too few taxa in characters and unaligned block
	- bug fix: only affects writing of NEXUS from file formats that do not
		contain tree names.  Trees were being printed without NEXUS names
	- validateInternalNodeLabels flag added to NxsTreesBlock. True by default, 
		but if set to false then the NxsTreesBlock will tolerate trees with
		internal labels that correspond to integers by storing the label as a
		string but not interpretting it as a taxon label
	- another example added: NEXUSsubsetter is not an install product.  It reads 
		a NEXUS file and a CharSet or TaxSet name



================================================================================
2.1.12 (relative to 2.1.09)
================================================================================

Removal of  trailing whitespace and large scale code reformatting will cause
	diff to indicate a huge number of changes in this release but the
	functionality has actually not changed much since 2.1.09

Bug fixes:
	- added more includes needed by g++ 4.4.01
	- NxsReader::ClearUsedBlockList is more sensible (it clears more fields).
	- NxsString no longer includel ncl/ncl.h it now includes ncl/nxsdefs.h
	- memory leak had been occurring when NxsExceptions were raised from a
		cloned block and the NxsReader was an ExceptionRaisingNxsReader
	- nxsdistancedatum.cpp and nxsemptyblock.* removed
	- block skipping -- now NCL will not be fooled by and end or endblock
subcommand.
	- now checks for continuous characters in IsGapState and IsMissing




Library enhancements:
	- much better documentation of v2.1 features
	- several convenience functions added especially to assist processing of the
		NxsCharactersBlock
	- added a modified version of NCLInterface.h from phylobase as an example.
		The file was written by Brian O'Meara and Derrick Zwickl
	- added more flexible, but admittedly hacky method for controlling the level
		of warnings that get printed out. Altering the warning level will also
		turn off status messages.  So a bool member was added to NxsReader so
		that client code can request that status messages still be shown.
	- added a Free method to Scoped... containers in nxsallocatematrix.h
	- added hot comments for autogeneration of v2.2 from v2.1 code

================================================================================
2.1.09 (relative to 2.1.06)
================================================================================
Bug fixes:
	- added includes needed by g++ 4.4.01

================================================================================
2.1.08 (relative to 2.1.06)
================================================================================
Bug fixes:
	- updated nexml writing to deal with changes in the schema
	- fixed bug causing mis-indexing of taxa in trees if there is no taxa block
		but a translate command.  This involves support for illegal nexus files.
		I believe that the bug would have showed diagnosable pathologies if it
		had affected any analysis (the highest number in the tree would have
been
		1 too high resulting in exceptions when the taxa block was queried).
	- bug fix when trees are followed by labels that contain tokens that have to
		be quoted
	- fixed behavior of GetNamedStateSetOfColumn and GetObsStates so that they
		are sensitive to the current gapmode for the char block
	- explicit inclusion of cstdlib where needed

Library enhancements:
	- Signal handling (via NxsReader::setNCLCatchesSignals ) commit r449
	- Lots of external format extensions and bug fixes:
		- converter application added in examples
		- export to AltNexus
		- CLUSTAL's .aln reading (not well tested)
================================================================================
2.1.06 (relative to 2.1.05)
================================================================================

Up to svn revision 376

bug fixes (visible to end-users):
	- opening files in binary mode (failing to do so cause problems with end-of-
		line and other special characters).
	- character index (0-based rather than 1-based number) was being printed in
		some contexts in error messages
	- Apple g++-4.2 optimization bug now has a workaround
	- Characters blocks now reject files in which the datatype is DNA, RNA, or
protein
		if SYMBOLS are added.


bug fixes API:
	- Fixed setting of NXS_HAS_ALL_TAXA_BIT
	- no longer duplicating blocks when the same reader is used to read more
		than one file
	- allowing PublicNexusReader::Add to be called
	- Some Windows-specific modifications to avoid warnings and errors compiling
		under Visual Studio 2008 (VC9)
	- characters block now generates an exception if there is no matrix
	- NxsReader::BlockIsASingeltonReader problem fixed
	- better cleaning up in the multiformat reader when there is an error in the
		reading of a file
	- another accessor added for NxsSimpleTree
	- PublicNexusReader GetXXXBlock functions changed to return 0L rather than
		raising an exception if you request a block that is out of range
	- NCL_CONST_FUNCS macro added

API additions:
	- improvements to phylip reading -- matchchar now accepted (multiformat
		reader only).
	- reading a single bare newick tree (as "phyliptree" or "relaxedphyliptree"
		in multiformat reader stuff).
	- added cullIdenticalTaxaBlocks() function to nexusreader.  You should call
		this before reading a file if you want all repeated taxa blocks to be
		destroyed rather than saved
	- support for genetic codes and translation to amino acids added
	- GetEscapedDouble and GetEscapedInt added
	- GetDatatypeMapForMixedType added
	- added functionality to delete identical taxa blocks even if they are
		implied blocks.  This only will be triggered if the client code has
		called NxsReader::cullIdenticalTaxaBlocks(true).  This corrects an
		issue DZ was seeing with reading multiple DATA blocks
	- aliasing of block titles for culled blocks
	- changed the handling of  the + character when a nxstoken instance has been
		altered by a call to indictate hyphenNotPunctuation is in effecet (this
		is done via the SetLabileFlagBit function). In almost all case this
		function is called so that a number (particularly one writtend in
		scientific notation will be read as a single token, rather than as many
		as 4 tokens).  In rare cases, the + is used before the exponent or even
		in front of the entire number.  So that client code does not have to
		deal with these cases, the default behavior has been changed to using
		the hyphenNotPunctuation to indicate that in the reading of the next
		token neither - nor + should be treated as punctuation.  Client code
		that wishes to use the old behavior can compile NCL with the
		NCL_VERSION_2_STYLE_HYPHEN macro defined to 1

================================================================================
2.1.05 (relative to 2.1.04)
================================================================================
Added include <cstdlib> at top of nxsdistancesblock.cpp and nxstoken.cpp to
avoid compilation errors on Fedora 9 systems.

================================================================================
2.1.04 (relative to 2.1.03)
================================================================================
Added inclusion of <climits> for gcc 4.3 (bug report by Brian Omeara)

================================================================================
2.1.03 (relative to 2.1.02)
================================================================================
minor bug fixes Jamroot file added for compilation  with bjam

================================================================================
2.1.02 (relative to 2.1.00)
================================================================================
minor bug fixes and const correctness

================================================================================
Since 2.1-specific changes.
================================================================================
Support for multiple instances of the same block type
	* NxsBlockFactory mechanism API - see the examples/normalizer for an
		example of usage. Full adoption of these API's is not required, but may
		make your life easier.
	* Reader now stores "implied blocks" -- e.g. makes a DATA block look to the
		client code like a TAXA block and CHARACTERS block. This improves the
		abstraction for library users.

Other features added:
	* commas tolerated in multiple state lists in the cells of a matrix {A,G}
		the same as {AG}.  Such files are not legal, but are common.
	* handling for TITLE and BLOCKID commands added to NxsBlock
	* optimizations (tokenizing and data structures in taxa block) -- now 50
		fold faster than 2.0 on the 11361 Taxon, 1360 Character file in the
		CIPRES test files.
	* TaxPartition and TreePartition read
	* SETS and CODONS blocks now read. Actually read as a big, fat ASSUMPTIONS
		block that spawns implied blocks (to deal with common practice of mixing
		the commands in these blocks).
	* WtSet, UserType, and TypeSet command
	* now tolerating DATATYPE not being first in FORMAT command (to deal with
		Clustal exports)
	* tokenizer now skipping embedded comments (note that this changes the
		behavior of command comment storing -- they are only returned if they
		precede token)
	* support for DATATYPE=MIXED added in CHARACTERS block
	* trees block can now validates trees as they are read to improve error
		reporting  -- this feature not used unless TreesBlock::
		SetProcessAllTreesDuringParse(true) is called on the trees block
		instance.
	* some support for NHX parsing -- in the TreesDescription.
	* support for LINK command (as introduced by Mesquite)
	* support Mesquite-style CharSet/TaxSet syntax.
 	* NCL_MAJOR_VERSION and  NCL_MINOR_VERSION defines added. You can wrap any
 		ncl-2.1 specific code (you shouldn't need much) as follows:

================================================================================
#	if defined(NCL_MAJOR_VERSION) && (NCL_MAJOR_VERSION >= 2) &&
(NCL_MINOR_VERSION >= 1)
		trees->SetAllowImplicitNames(true);
#	endif
================================================================================



bug fixes:
	* RNA equates bug reported by DZ
	* case of equate keys  (characters of both cases are now added for molecular
		data) reported by DZ
	* Tax and Charset reading now tolerate Standard keywords (they still
		generate errors for VECTOR)
	* fixes to the Distances block if the input order is not identical to the
		taxa block order
	* Some CharSet reading
	* Lots more that I forgot to document.

Behavior changes (could be called features or bugs depending on your POV):
	* Now converts any built in CHARACTERS datatype that has symbols extensions
		to standard datatype.



================================================================================
Since 2.0.0
================================================================================
12 Dec 2007 Mark Holder (mholder@users.sourceforge.net)
	* Version 2.0.04 (https://svn.sourceforge.net/svnroot/ncl/branches/v2-stable
revision 75)
	* new features: Basic handling of Continuous datatype (ITEMS now
		supported).  Brian O'Meara contribute code and suggestions to this
		new feature.
	* bug fixes: taxa information after a treedef (root taxa info) is now
		accepted

30 Nov 2007 Mark Holder (mholder@users.sourceforge.net)
	* Paul's unaligned parsing block functionality added
	* bug fixes (newtaxa is now implicit if ntax is used, other small changes
		mainly pertaining to writing NEXUS, and testing with a round trip)


18 Dec 2006 Mark Holder (mholder@users.sourceforge.net)
	* Version 2.0.02 (https://svn.sourceforge.net/svnroot/ncl/branches/v2-stable
revision 47)
		- bug fixes to Version 2.0.01 (errors introduced in CharactersBlock
Version 2.0.01 while suppressing errors)
		- improvements to regression testing

03 Nov 2006 Mark Holder (mholder@users.sourceforge.net)
	* Version 2.0.01
		- minor reorganization (examples dir)
		- suppressing warnings
		- reworked Makefile and configure (fixes link error from multiply
defined objects on some Linux variants)

20 May 2003 Paul Lewis (paul.lewis@uconn.edu)
	* Version 2.0 released (first official release)

================================================================================
Latest changes
================================================================================