= 2.3.0

This release adds I/O features.

Seeking forward to an arbitrary offset is achieved with Skip's :to_abs_offset
parameter.

Multi pass I/O has been added.  BinData performs I/O in a single pass.  See
DelayedIO to perform multi pass (backwards seeking) I/O.

The experimental :adjust_offset feature has been removed.  If you have existing
code that uses this feature, you need to explicitly require 'bindata/offset' to
retain this functionality.

= 2.1.0

Several new features in this release.

* Fields can be aligned on n-bytes boundaries.  This make it easier to write
  parsers for aligned structures.
* The endian value :big_and_little can be used to automatically create :big and
  :little endian versions of the class.
* Bit fields can be defined to have their length determined at runtime.
* The state of :onlyif fields can be determined by #field?

== Deprecations

The #offset method has been renamed to #abs_offset.  This make an obvious
distinction to #rel_offset.

= 2.0.0

BinData 2.0.0 requires ruby 1.9 or greater.  Support for ruby 1.8 is found in
BinData versions 1.8.x.

Prior to version 2.0.0, BinData used the Ruby 1.8 conversion of strings for
method names.  As of 2.0.0, BinData uses symbols.

    class A < BinData::Record
      uint8 :a
      uint8 :b
    end

    # BinData 1.8.x
    A.read "\001\002" #=> {"a"=>1, "b"=>2}

    # BinData >= 2.0.0
    A.read "\001\002" #=> {:a=>1, :b=>2}

= 1.6.0

Added :assert as a replacement for :check_value.  Note that :assert performs
the checking on assignment as well as reading.

The parameter :asserted_value is a new shortcut for combining :assert and :value
    int8 :magic, :assert => 42, :value => 42

Can be written more concisely as
    int8 :magic, :asserted_value => 42

= 1.5.0

Finally moved the source code to github.

Deprecated functionality has been removed to cleanup the codebase.

= 1.4.4

== Deprecations

The BinData::String parameter :pad_char has been renamed to :pad_byte.  This
aids readibilty as Ruby 1.9 characters are not necessarily one byte.

= 1.4.0

== Deprecations

There is no longer any need to call #register_self when inheriting from
BinData::Base.

BinData::Wrapper is deprecated.  You should use direct inheritance instead.

If your code looks like:
    class Uint8Array < BinData::Wrapper
      default_parameter :initial_element_value => 0
  
      array :initial_length => 2 do
        uint8 :initial_value => :initial_element_value
      end
    end

It should be changed to:
    class Uint8Array < BinData::Array
      default_parameter :initial_element_value => 0
      default_parameter :initial_length => 2
  
      uint8 :initial_value => :initial_element_value
    end

See http://bindata.rubyforge.org/manual.html#extending_existing_types for details.

= 1.3.0

== New features

You can now assign values to BinData objects when instantiating them.

Previously:
    obj = BinData::Stringz.new(:max_length => 10)
    obj.assign("abc")

Now:
    obj = BinData::Stringz.new("abc", :max_length => 10)

== Backwards incompatible changes

This version makes some backwards incompatible changes for more advanced
users of BinData.

This only affects you if you have created your own custom types by
subclassing BinData::Base or BinData::BasePrimitive.

All instance variables must now be initialized in #initialize_instance.
Implementing #initialize is no longer allowed.

Run your existing code in $VERBOSE mode ("ruby -w"), and BinData will
inform you of any changes you need to make to your code.

If your code previously looked like:

    class MyType < BinData::Base
      register_self

      def initialize(parameters = {}, parent = nil)
        super

        @var1 = ...
        @var2 = ...
      end
      ...
    end

You must change it to look like:

    class MyType < BinData::Base
      register_self

      def initialize_instance
        @var1 = ...
        @var2 = ...
      end
      ...
    end

= 1.0.0

Version 1.0.0 removes all deprecated features.  If you are upgrading from a
version prior to 0.10.0 you will need to make changes to your code before it
will work with BinData 1.0.0 or later.

It is recommended to first upgrade to 0.11.1, and run your code with the -w
command line switch.  This will turn on warning messages that should give you
hints about which parts of your code need changing.  See the NEWS for 0.10.0
for details of the deprecated features.

= 0.10.0

There are several new features in this release.  The major ones are:

* Debugging declarations is now easier with the ability to trace values when
  reading.

      BinData::trace_reading(STDERR) do
        obj_not_quite_working_correctly.read(io)
      end

  Will result in a debugging trace written to STDERR.

* Support for arbitrary sized integers and bit fields.

* Struct/Array field/element access now returns a BinData object, rather than
  the underlying Fixnum or String.  The BinData object will behave like it's
  underlying primitive so existing code should continue to work.  This allows
  for an improved syntax in your declarations.

  If your code requires access to the underlying primitive, you can access it
  with the #value method.

There are several deprecations and one backwards incompatible change in this
release.  Turn on $VERBOSE mode (-w command line switch) to see warnings
regarding these deprecations.

== IMPORTANT - Ruby does not warn you about this change!

* The #to_s method for getting the binary string representation of a data
  object has been renamed to #to_binary_s.  If you use this method you must
  manually change your code or you will get strange results.

== Deprecations.  Ruby will warn you of these when in $VERBOSE mode.

Code using these deprecated features will still work in this release but will
fail to work in some future release.  Change your code now.

* BinData::SingleValue has been renamed to BinData::Primitive.

* BinData::MultiValue has been renamed to BinData::Record.

* String :trim_value parameter has been renamed to :trim_padding.

* Registry.instance should be replaced with RegisteredClasses.

* struct.offset_of("field") should be replaced with struct.field.offset.

* struct.clear("field") should be replaced with struct.field.clear.

* struct.clear?("field") should be replaced with struct.field.clear?.

* struct.num_bytes("field") should be replaced with struct.field.num_bytes.

* array.clear(n) should be replaced with array[n].clear.

* array.clear?(n) should be replaced with array[n].clear?.

* array.num_bytes(n) should be replaced with array[n].num_bytes.