### Version 1.1.0 / 2016-10-28

* Support added for non-English dialects. This gem should now be able to model 
  feature files using any dialect supported by the 'gherkin' gem.

* Models for elements of Gherkin that have keywords (e.g. 'Feature', 'Scenario', 
  'Examples') now keep track of the keyword used by the element that they model.

* Bug fix - Fixed a bug that was causing example models to output extra newline 
  characters under certain circumstances. 


### Version 1.0.4 / 2016-10-07

* Bug Fix - Fixed a bug that caused some models to include nil objects in their 
  children collection if they did not have the relevant child object.


### Version 1.0.3 / 2016-09-12

* Fixed a gem dependency that was accidentally declared with '<=' instead of '<'.


### Version 1.0.2 / 2016-09-12

* A more detailed gem description and summary have been added to the gemspec.

* The gem now declares version limits on its dependencies.
 
* Badges for the current status of the project have been added to the Readme.


### Version 1.0.1 / 2016-09-10

* In the Readme file, added a link to the published documentation.


### 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