Version 2.0.0

  * New cardinality calculator. This improves support for schemas that
    use complex structures with repeated elements, e.g.,

    <complexType name="Type">
      <choice>
        <sequence>
          <element name="a" type="string"/>
          <element name="c" type="string"/>
        </sequence>
        <sequence>
          <element name="b" type="string"/>
          <element name="c" type="string"/>
        </sequence>
      </choice>
    </complexType>


  * New identifier escaping code. With this feature xsd generates proper
    code for schemas that use the same name for an element and an attribute
    in the same type or use several elements/attributes with different
    qualified names but with the same local name, e.g.,

    <!-- base.xsd -->
    <schema xmlns="http://codesynthesis.com/xmlns/test/foo"
            targetNamespace="http://codesynthesis.com/xmlns/test/foo">

      <element name="foo" type="int"/>
    </schema>

    <schema xmlns="http://codesynthesis.com/xmlns/test/bar"
	    xmlns:f="http://codesynthesis.com/xmlns/test/foo"
            targetNamespace="http://codesynthesis.com/xmlns/test/bar">

      <import namespace="http://codesynthesis.com/xmlns/test/foo"
              schemaLocation="base.xsd"/>

      <element name="foo" type="string"/>

      <complexType name="Foo">
        <sequence>
          <element ref="foo"/>
          <element name="foo" type="long"/>
          <element ref="f:foo"/>
          <element ref="f:foo"/>
        </sequence>
        <attribute name="foo" type="string"/>
      </complexType>
    </schema>


  C++/Tree

    * New option, --generate-polymorphic, triggers generation of
      polymorphism-aware code. Before this release xsd always used to
      generate polymorphism-aware code. However, it appears to be quite
      wasteful in terms of the generated code size (up to 40%). You will
      now need to explicitly specify this option if you use substitution
      groups or xsi:type. A warning is issued if this option is not
      specified but the schema makes use of substitution groups.

    * New options, --root-element-first, --root-element-last,
      --root-element-all, --root-element-none, and --root-element, control
      generation of parsing and serialization functions. With these options
      you can avoid generating extra code for global elements that are not
      document roots. See man pages for details.

    * New options, --parts and -parts-suffix, allows you to split generated
      source code into a number of parts. This is useful when translating
      large, monolithic schemas and a C++ compiler is not able to compile
      the resulting source code at once (usually due to insufficient memory).

    * New option, --generate-default-ctor, triggers generation of default
      constructors even for types that have required members. Required
      members of an instance constructed using such a constructor are not
      initialized and accessing them results in undefined behavior. Thanks
      to Jean-Francois Dube <jf at magnu.polymtl.ca> for suggesting this
      feature.

    * New option, --generate-from-base-ctor, triggers generation of
      constructors that expect an instance of a base type followed by all
      required members. Thanks to Jean-Francois Dube <jf at magnu.polymtl.ca>
      for suggesting this feature.

    * Information scopes for attributes and elements with default/fixed values
      now define the public static default_value function which allows one to
      obtain the default/fixed value for the element/attribute. Thanks to
      Dave Moss <david.r.moss at selex-comm.com> for suggesting this feature.

    * MSVC 7.1 has a limit on the length of the "if else if" chain. This
      results in ICE when compiling generated code for enumerations with
      a large number of values. This version addresses this issue. Thanks
      to Cyrille Chplov <cyrille at chepelov.org> for reporting this and
      suggesting a fix.


  C++/Parser

    * The parser construction API has changed. Now, for element 'foo',
      the name of the parser modifier function is 'foo_parser'. Likewise,
      operator() for setting all parsers at once has been changed to the
      'parsers' function.


Version 1.9.0

  C++/Tree

    * The size modifier function in the base64_binary and hex_binary
      built-in types automatically adjusts capacity if needed.

    * More internal names (names that start with _xsd_) were made
      private or protected.

  C++/Parser

    * Typedef for the parser base in the xml_schema namespace.

  C++/Parser-E

    * C++/Parser mapping optimized for embedded systems. For now it
      is equivalent to 'cxx-parser --xml-parser expat'.


Version 1.8.0

  * Moved to the build 0.2 series.

  C++/Tree

    * Support for default and fixed values in attributes. An optional
      attribute with a default or fixed value is mapped to the One
      cardinality class instead of the Optional cardinality class.

    * Mapping for base64Binary and hexBinary has improved. Now these
      types support a basic buffer abstraction and perform automatic
      encoding and decoding.

    * Internal names are protected. We've noticed (via bug reports) a
      wide use of internal names (names that start with _xsd_) in user
      code. This is not portable and instead you should use public
      names. To prevent this from happening in the future we've made
      all internal names protected.

  C++/Parser

    * Support for Expat as the underlying XML parser in addition to
      Xerces-C++. This allows one to use the C++/Parser mapping in
      memory-constrained environments such as embedded systems. To
      select Expat instead of Xerces-C++ (default) add
      '--xml-parser expat' to the command line. At the moment only
      'char' (UTF-8) is supported as the base character type when
      Expat is selected.

    * The invalid_instance exception has been renamed to parsing.

    * Generic error_handler interface has been added in addition
      to Xerces-C++-specific DOMErrorHandler. It allows you to
      handle parsing errors and warnings without having to deal
      with Xerces-C++ specifics.

    * The default error handling behavior has changed in parsing
      functions. Instead of printing errors and warnings to STDERR,
      the errors are now collected and thrown as part of the parsing
      exception.

    * In parsing functions, the name, namespace arguments order has
      been reversed to be consistent with the one used in parsing
      hooks.

Version 1.7.0

  * Number of bug fixes in libxsd and the generated code.

  C++/Tree

    * Comprehensive XML Schema C++/Tree Mapping User Manual.

    * Basic support for union. A simple type that is defined using
      derivation by union is mapped to a C++ class that derives from
      string.

    * The _clone function has its arguments default-initialized.

    * The invalid_instance exception has been renamed to parsing.

    * Generic error_handler interface has been added in addition
      to Xerces-C++-specific DOMErrorHandler. It allows you to
      handle parsing/serialization errors and warnings without
      having to deal with Xerces-C++ specifics. See the user
      manual for more information.

    * The default error handling behavior has changed in parsing
      and serialization functions. Instead of printing errors and
      warnings to STDERR, the errors are now collected and thrown
      as part of the parsing/serialization exception. See the user
      manual for more information.

    * The optional and sequence containers now support operators ==,
      !=, <, >, <=, and >=.

    * Flags argument has been added to serialization functions. The
      only flag that is currently supported is dont_initialize.

    * Generated code cleanups.

  C++/Parser

    * Basic support for union. A simple type that is defined using
      derivation by union is mapped to a C++ class template that
      is just an alias for the generic parser. You are expected to
      override the _characters function in your implementation.

    * Properties argument to parsing functions which allows to
      programmatically specify schemas for instance document
      validation.

    * Flags argument to parsing functions. The following flags
      are supported:

      dont_validate   - do not validate instance documents
      dont_initialize - do not initialize the Xerces-C++ runtime

Version 1.6.0

  * Number of bug fixes in libxsd and the generated code.

  C++/Tree

    * Support for xsi:type and substitution groups in parsing and
      serialization. See examples/cxx/tree/polymorphism for a code
      sample.

    * Properties argument to parsing functions which allows to
      programmatically specify schemas for instance document
      validation.

    * Extra checks in parsing code which prevents construction
      of inconsistent in-memory representation from invalid
      instance documents. Should be useful when validation is
      disabled.

    * Accessors and modifier were made normal member functions.
      Before they were implemented via functors.

    * Workaround for g++-3.3 bug# 16650:

      http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16650

  C++/Parser

    * All "service" functions were renamed to start with '_'.
      This should prevent hiding of service functions by
      elements/attributes with the same names.

Version 1.5.0

  * Number of bug fixes in libxsd and the generated code.

  C++/Tree

    * Basic support for inheritance-by-restriction in complex types.

    * The following parsing flags have been introduced:

      keep_dom        - keep association with underlying DOM nodes
      dont_validate   - do not validate instance documents
      dont_initialize - do not initialize the Xerces-C++ runtime

    * "Type-less content" such as mixed content models, xsd:anyType/
      xsd:anySimpleType, and xsd:any/xsd:anyAttribute is supported by
      exposing corresponding DOM nodes (see the keep_dom parsing flag).
      Note that only a subset of XML Schema xsd:any functionality is
      supported. The compiler will issue diagnostics for unsupported
      cases. See examples/cxx/tree/mixed for a code sample.

  C++/Parser

    * Support for inheritance-by-restriction in complex types.

    * "Type-less content" such as mixed content models, xsd:anyType/
      xsd:anySimpleType, and xsd:any/xsd:anyAttribute is supported
      by forwarding parsing events to a set of "unexpected" hooks.
      Note that only a subset of XML Schema xsd:any functionality is
      supported. The compiler will issue diagnostics for unsupported
      cases. See examples/cxx/parser/mixed for a code sample.

Version 1.4.0

  * Number of improvements and bug fixes in the diagnostics code.

  * libxsd has been reorganized to provide a clean split of code with
    regards to char/wchar_t use. It should be possible to use libxsd
    and the xsd-generated code on platforms that lack wchar_t support,
    such as mingw.

  C++/Tree

    * Work around for g++ bug# 23206.

    * Support for xsd:list.

    * Type/member name conflicts are auto-resolved. Such conflicts
      occur when a type and an element or attribute withing this type
      share the same name.

    * XML Schema extension, the 'refType' attribute, allows one to
      specify referenced type for xsd:IDREF and xsd:IDREFS data types.
      See examples/cxx/tree/library for details.

    * New option, --morph-anonymous, allows automatic morphing
      of anonymous types to named ones. See the man pages for
      details.

    * New option, --namespace-regex-trace, allows one to trace the
      namespace mapping process. See the man pages for details.

    * Mapping for optional elements/attributes (cardinality 0..1)
      has changed in a backwards-incompatible way. In the previous
      version you would write:

      Bar& bar = ...

      if (bar.foo.present ())  // test
      {
        Foo& foo (bar.foo ()); // get

        bar.foo (Foo (...));   // set

        bar.foo.reset ();      // reset
      }

      Now you would write it like this:

      if (bar.foo ().present ())      // test
      {
        Foo& foo (bar.foo ().get ()); // get

        bar.foo (Foo (...));          // set

        bar.foo ().reset ();          // reset
      }

      Or using the pointer notation:

      if (bar.foo ())           // test
      {
        Foo& foo (*bar.foo ()); // get

        bar.foo (Foo (...));    // set

        bar.foo ().reset ();    // reset
      }

  C++/Parser

    * Support for xsd:list.

    * Type/member name conflicts are auto-resolved. Such conflicts
      occur when a type and an element or attribute withing this type
      share the same name.

    * New option, --namespace-regex-trace, allows one to trace the
      namespace mapping process. See the man pages for details.

Version 1.3.0

  * Numerous bug fixes.

  * The XML subsystem of libxsd has been reorganized to provide
    a clean split of DOM and SAX functionalities.

  C++/Parser

    * New option, --morph-anonymous, allows automatic morphing
      of anonymous types to named ones. See the man pages for
      details.

  C++/Tree

    * Additional parser functions provide support for reading
      from std::istream.

Version 1.2.0

  C++/Parser

    * New backend that generates the C++/Parser mapping.

Version 1.1.1

  all backends

    * Bug fixes in the filesystem path handling logic.

Version 1.1.0

  C++/Tree

    * New option, --generate-serialization, triggers generation of
      serialization functions. Serialization functions convert an in-memory
      representation back to XML.

    * xsd::cxx::tree::vector has been extended to closely follow std::vector
      API. This allows you to access and modify element sequences as if they
      were of type std::vector.

    * Generated constructors from xml::attribute and xml::element are made
      explicit.

    * The library example was extended to showcase modification and
      serialization of the in-memory representation.

    * New "XML Schema C++/Tree Mapping Serialization Guide" has an in-depth
      treatment of the serialization mechanisms provided by xsd.

Version 1.0.1

  all backends

    * Improved diagnostics.

    * Bug fixes in the schema inclusion/importing logic.

  C++/Tree

    * Two new options: --include-with-brackets and --include-prefix

Version 1.0.0

  * First public release.
