Quickcheck 0.5 release

It's done: Quickcheck version 0.5 is ready. Again one of the nice little releases of Quickcheck.

The 0.5 release added the following features mainly to provide better readability:

• Support for an adapter from Generator to Iterable to allow the use of Java's for-expression in tests (replacing the inner classes).
• Generated source code for @Iterables and @Samples annotations on generator factories. The @Iterables annotation was added to support for-expression directly with Generators without additional code. The @Samples annotation can be used to generate a factory for sample values from a Generator class.
  

Minor changes are:

• Added a declarative Pojo generator for interfaces.
• Implemented excluding generator.
• Added Generators.cast utility method. Generator<String> can now be casted to Generator<Object> to provide broader applicability of Generators.
• Maven repository support (http://download.java.net/maven/2/net/java/quickcheck)
  

The version 0.6 of Quickcheck will be a major rework after more than 2 years of experience with the specification-based test approach and development. Mainly to support further development for example shrinking (simplification) of failed test values. A major feature still missing in Quickcheck version 0.5.

When I started Quickcheck I would not have thought that it is possible to develop such a tiny little thing for more than 2 years. Still more ideas and things that could be implemented and written about and so little time.

Recently, I've used the 0.5 version in my Scala projects I blogged about. I did not mention that I tested it in any way, though. It was a pleasant experience given that QuickCheck was not designed with Scala's language features in mind. It works with just a small implicit type conversion and adaption layer (10 lines or so).

Distribution function implementations based on the XorShift random number generator I've already worked with would be nice to have. When I'm at adding new distributions anyway, there are still the distribution functions based on heuristics to implement. For example a distribution function based on a boundary heuristic that starts production of values with minimum value, -epsilon, 0, +epsilon and maximum value before random generation kicks in. Another item on my ever growing list is the description of patterns related to specification-based testing and QuickCheck. Obviously there is still plenty of work left.