=== Version 1.0.0 / 2016-09-08

* All modeled elements of a Cucumber test suite are now modeled with actual classes. Previously, some elements (such as
  tags or rows in an example table) were modeled with simple strings.

* A base model class has been added in order to simplify adding new element models. Other classes and modules used for
  organizing common model behavior have been created/removed/renamed.

* Feature file models now only have a single feature model instead of a collection of them (that would only ever have
  one item in it).

* Rows in an example table and rows in a step table no longer use two different classes for modeling.

* Background models can now be compared to other models that have steps (i.e. scenarios and outlines).

* All models for Gherkin elements now track the line number from which they originate in a source file.

* The source line of a model is now a mutable attribute instead of being read only.

* The file path of a feature file is now a mutable attribute instead of being read only.

* Standardized the initial values of the attributes of abstractly created models so that they are consistent across
  model types.

* Added specific ancestor types for scenario, outline, and background models as alternatives to the generic 'test'
  ancestor type.

* Extra whitespace around parsed element descriptions is now trimmed away before being stored in a model.

* Adding rows to examples via hashes can now be done without regard to the order of the keys in the hash. The correct
  order can be determined by the model.

* The saved parsing data that is generated by the 'gherkin' gem is no longer duplicated for each model, resulting in
  significantly less memory usage.

* String output of models has been improved. More special characters in Gherkin (e.g. vertical bars in rows) are
  appropriately escaped in the string output of a model and several minor bugs related to using model string output as
  the input text for new models have been fixed.

* Removed the backdoor used to pass around model data during the model creation process. Model input can now only be
  text.

* Simple counting methods have been removed. Ruby's collection methods are just as easy to use directly when dealing
  with classes that contain collections and the removal of these counting methods simplifies the codebase.

* The 'World' module has been removed. The scope of this gem is to model files written in Gherkin. Concepts like
  'step definitions' and 'hooks' are better left to libraries that include (possibly programming language specific)
  test execution logic within their scope of concern.

* Step models no longer have 'arguments' because they require the concept of step definitions, which are no longer
  within the scope of this gem.

* Removed deprecated behavior
   - Marked
       - Doc strings no longer use an array of strings to model their content
       - Tables no longer use nested arrays of strings to model their rows
       - Models that have descriptions no longer use an array of strings to model their description
       - The convenient (read: awkward) #step_text has been removed.
   - Unmarked
       - Examples no longer use an array of hashes to model their rows

* Various methods have been renamed.


=== Version 0.4.1 / 2016-05-12

* Increased the flexibility of input when adding rows to an Example object. Non-string values can now be used as input
  and they will be converted into strings. Original input objects are not modified.
* Added some error checking around adding value rows to an Example object without adding a parameter row as well.


=== Version 0.4.0 / 2016-05-01

* The path of a Directory object is now a changeable attribute instead of only being populated if the instance was
  given a diretory to model.
* Bug fix - Fixed a bug that occurred if a Directory object was asked for its #name when it was created as 'abstract'
            instead of modeling an existing directory.


=== Version 0.3.0 / 2016-04-24

* Support for version 4.x of the 'gherkin' gem added.
* Bug fix - Fixed a bug that was preventing Example objects from being created from text if that text had less Gherkin
            structure than normal.


=== Version 0.2.0 / 2016-02-21

* Better error feedback when parsing errors are encountered. It is now easier to tell which file contained invalid
  Gherkin.


=== Version 0.1.0 / 2016-02-10

* Support for version 3.x of the 'gherkin' gem added.
* Bug fix - The saved parsing data that is generated by the 'gherkin' gem is no longer modified by the rest of the
            model creation process.


=== Version 0.0.2 / 2015-11-22

* Bug fix - Fixed a bug that was causing object comparison using #== to not work when comparing some models to other
            types of objects.


=== Version 0.0.1 / 2014-06-02

* Initial release