* *    PLUGIN END-OF-LIFE NOTICE
* *   
* *    This plugin is now part of Edge Rails and there is no further development planned.
* *    If you are running Rails Edge or Rails at or above 1.2 you don't need to use this plugin!
* *    Thanks everyone.
* *    Julik signing off.


============================================
* Parameter normalization should not damage non-string immediate values such as integers (julik)
* If you define an env variable called AS_MULTIBYTE_SKIP_HANDLING_TESTS (duh!) UTF-8 handling tests will be skipped (julik)
* Pure2 handler now does grapheme cluster length instead of codepoints length (manfred)
* Updated the pure2 handler tables to unicode 5.0.0 (manfred)
* Fixes the Normalization plugin to use Controller#params (julik)
* Urgent fixes to cope with the new Dependency code from edge (julik)
* Changes to the docs - uppercase French withouth accents is not trendy (julik)
* Added info on the version of the standard (julik)
* Fixed assertions for upcase so that we lose composed ligatures and ßuch (julik)
* Excising extraneous normalization code from filters and preventing an exception (julik)
* Minor changes (julik)
* Non-match should return nil (julik)
* Fixed the test imports so they work in rails 1.1.4 (manfred)
* Actually move the pure2 generator. Fixed paths in the script (manfred)
* Renaming normalize_params to normalize_unicode_params (julik)
* Normalizing filter will skip normalization for improper UTF sequences (julik)
* Made changes to the normalization filter (julik)
    * Normalization filter is a class method, as per docs
    * Accepts the :form option
    * Default form option is KC
    * Bails on unknown options
    * The Chars has normalize and normalize! that accepts the form parameter
    * The controller normalization is inheritable, and will define the accessor for form on the controller.
    * Defining the normalize_params setting many times will only override the form, not duplicate the filter
    * Added module-wide parameter DEFAULT_NORMALIZATION_FORM, used all over the place
    * Moved normalization filter into a potential ActionPack extraction
* More changes to the PRIMER and reformatted it for fixed length (please don't reflow!) - (julik)
* Moved stuff around to adhere to ActiveSupport and preparing the merger (julik)
    * Table generator moved to multibyte/generators
    * Tables moved to values
    * Performance script moved to test
* Small changes to README and UNICODE_PRIMER (manfred)
* Integrated LICENSE into Rdocs (julik)
* Added a license file (manfred)
* Removed implicit normalization everywhere (manfred)
* Small corrections to the README (julik)
* Removed a level of redirection (handler_for) because we only have utf-8 handlers right now (manfred)
* The multibyte handler is now kept in a class variable (manfred)
* Change defaults in the {upper,lower}case_mapping to speed up upcase/downcase (manfred)
* Every handler now has a 'decompose' method which does NFD. (manfred)
* Removed test for 'Unicode' loading, we have pure ruby normalization (manfred)
* Added new pure ruby handler (manfred)
* Added normalize_KC methods in utf8_handler.rb where the documentation and the tests said it should normalize: reverse and size (manfred)
* utf8_handling_test renamed, it can now be run seperately (manfred)
* Quick block to skip normalization tests to shorten testing (julik)
* Added explicit bypass to gsub to Chars, because each object has gsub from Kernel (julik)
* Fixed small spelling and documentation errors (manfred)
* Fixed the t_controller tests; you can't have more that one post, get or xhr in one test (manfred)
* Made ActiveSupport::Multibyte::StringExtensions#u an alias for #chars instead of a method (manfred)
* Handler implemented based on UTF8Proc (julik)
* Return an array of Chars from #split (julik)
* We have a pure Ruby normalizer backported from utf8_proc (thx meadow.nick)
* TEMPORARY swappable Handlers for testing, to test agains Unicode conformance cases and benchmarking (julik)
* Added tests from Unicode Consortium to the testing harness - downloaded when the test is ran (julik)
* Began work a testing harness to check Unicode implementations (julik)
* Whoops, had to return explicitly (julik)
* Testing checkin from fingertips (julik)
* Calling setup_unicode_client only if it's available, SQLite adapter does not inherit from abstracts (thx David Marko)  (julik)
* Changed regexes for leaders and trailers to anchor properly to strings instead of lines (julik)