===> 0.2.1:

--- 08/31/2003

Found a major typo in the docs regarding parsing flags.  Can't have that.

===> 0.2.0:

--- 08/31/2003

Added quixml::create_node which creates a node as a hash which contains all of
the expected properties (name, attributes, children, text).  Strictly for
convenience.

Added example code to every method in the docs.

Change the "programming with" section in the docs to be more exemplary and less
lecturing.

--- 08/30/2003

quixml::parse, quixml::tree.parse and quixml::tree.to_xml all now take an
optional parameter which is a flag of bitwise values which control how XML is
parsed and output.

Output didn't quite adhere to its own rules; fixed that.

--- 08/29/2003 sdodell

More distribution fine-tuning.  Better version reporting on the home page,
generally more automated.

Added find_content method.

Changed named of find_content to content_of

find_all now calls a block if given.

Renamed internal functions to better differentiate between class and instance
methods.

Added class method parse to create a tree directly from xml.  It calls a block
same as parse.

Added class method with_node which uses a node as the root for a new
quixml::tree and passes the tree as the parameter to a block.

Added attributes_of and children_of and changed content_of to text_of.

The term "content" is replaced completely with the term "text" everywhere.

When element addressing is performed, the value "." automatically always
matches the root node now.

More internal cleanups; renaming of things here and there.

Added constants for better control over parsing and printing regarding extra
whitespace, data mixed with elements and tag appearance.

===> 0.1.1:

--- 08/29/2003 sdodell

Fixed the distribution build so it decompresses in its own directory.

===> 0.1.0:

--- 08/28/2003 sdodell

Tuned makedist.rb

Completely re-analyzed how memory was being handled and changed just about
everything related to garbage collection. First, since no Ruby values or
memory blocks were being kept by QuiXML::Tree, removed all rb_gc_reg/unreg
type of calls; theoretically, the garbage collector should know how to handle
all that, so I took my hands off it all.  Then, I changed all the structures
that were being created on the stack to be created by ALLOCA_N, which will
"mark" any values or memory blocks kept by the structures.  There was a bug
where the garbage collector was collecting memory I had allocated with malloc
because it was being allocated in a stack frame deeper than the caller, and
was getting unmarked.

In a nutshell, memory should be very well taken care of now.

--- 08/27/2003 sdodell

Switched to Ruby-allocated memory; that should stop any potential leaks.

Fixed a problem where the wrong variable holding a node value was registered
with the GC.

Fixed a crash bug caused by a global variable whose storage fell off the stack.

Fixed the whitespace stripping.