NEWS

Version 2.5.0

 * Database classes are now move-constructible. This means they can be
   returned by value from functions in C++11 and later.

 * Support for mapping C++ types as other C++ types. For example:

   #pragma db map type(bool)                 \
                  as(std::string)            \
                  to((?) ? "true" : "false") \
		  from((?) == "true")

   See Section 14.8.1, "C++ Type Mapping Pragmas" in the ODB manual for
   details.

 * Support for custom table definition options in addition to column
   definition options. For details, refer to Section 14.1.16, "options" in
   the ODB manual.

 * New helper header, <odb/nested-container.hxx>, provides manual nested
   container support. For details, see the container-nested example in
   odb-examples.

 * New helper header, <odb/c-array-traits.hxx>, provides optional
   mapping of C arrays as containers. Note that this mapping is not
   enabled by default. See instructions inside the header for details.

 * Support for nested object ids. Now the `id` pragma specifier can optionally
   include the data member path to the id inside a composite value. For
   example:

   #pragma db id(first)
   std::pair<int, int> p;

   Note that one somewhat counter-intuitive aspect of this new feature is
   that the whole member marked with id (`p` in the above example) and not
   just the actual id member (p.first in the above example) is treated as
   read-only.

   Such nested id also cannot be automatically assigned (that is, use the
   `auto` specifier).

   To match this support the `inverse` pragma specifier now also allows
   nested data members.

 * Support for defining views as instantiations of C++ class templates,
   similar to objects and composite value types.

 * Support for using object pointers as map keys. Also the restriction for map
   keys and set values to be NOT NULL was removed.

 * New `points_to` pragma allows the establishment of relationships without
   using object pointers. See Section 14.4.37, "points_to" in the ODB manual
   for details.

 * A statement in a view that is prefixed with the /*SELECT*/ comment is
   recognized and handled as a SELECT statement. This can be used to work
   around unrecognized SELECT query prefixes.

 * Support for ordering virtual data members. For details, see Section
   14.4.13, "virtual" in the ODB manual.

 * The special (!) placeholder denotes the database instance in the modifier
   expressions. For details, see Section 14.4.5, "get/set/access" in the
   ODB manual.

 * Support for bulk operations in PostgreSQL 14 using the new pipeline mode.
   For details on bulk operations see Section 15.3, "Bulk Database Operations"
   in the ODB manual. Note that while this functionality requires libpq
   version 14 or later, it should be usable with PostgreSQL servers version
   7.4 or later. The development of this support was sponsored by Qube
   Research & Technologies Limited.

 * Support for calling PostgreSQL stored procedures and functions. For
   details and limitations refer to Section 19.7, "PostgreSQL Stored
   Procedures and Functions" in the ODB manual.

 * New serial_connection_factory in the SQLite runtime.

   This factory can be used when the database access is guaranteed to be
   serial. See Section 18.3, "SQLite Connection and Connection Factory"
   in the ODB manual for details.

 * Support for attaching additional SQLite databases to the main database
   connections (ATTACH DATABASE statement). See Section 18.4, "Attached
   SQLite  Databases" in the ODB manual for details.

 * Support for SQLite incremental BLOB/TEXT I/O (the sqlite3_blob_open()
   functionality). For details, refer to Section 18.1.3, "Incremental
   BLOB/TEXT I/O" in the ODB manual.

 * Support for mixed auto/manual id assignment in SQLite.

   Now one can do:

   #pragma db id auto
   odb::nullable<int64_t> id;

   And then set the id to NULL to get auto-assignment or to the actual value
   to use a manual id.

 * Support for mixed auto/0 id assignment in MySQL.

   Now one can do:

   #pragma db id auto
   odb::nullable<int64_t> id;

   And then, when used with NO_AUTO_VALUE_ON_ZERO, set the id to NULL to get
   auto-assignment or to 0 to use 0 as the id.

 * Map string object ids to MySQL VARCHAR(128) instead of 255 to support
   4-byte UTF-8.

   This is a backwards-incompatible change in that it may change your schema.
   To obtain the old behavior you will have to explicitly re-map std::string
   with the id_type pragma or explicitly specify the database type for each
   affected id member with the type pragma.

 * Due to warnings issued by some compiler, std::auto_ptr is no longer mapped
   as an object pointer or wrapper in the C++11 and later modes. Use
   std::unique_ptr instead.

Version 2.4.0

 * Support for object loading views. Object loading views allow loading of
   one or more complete objects instead of, or in addition to, a subset of
   data members and with a single SELECT statement execution. For details,
   refer to Section 10.2, "Object Loading Views" in the ODB manual.

 * Support for bulk operations in Oracle and SQL Server. Bulk operations
   persist, update, or erase a range of objects using a single database
   statement execution which often translates to a significantly better
   performance. For details, refer to Section 15.3, "Bulk Database
   Operations" in the ODB manual.

 * New database class functions, query_one() and query_value(), provide
   convenient shortcuts for situations where the query is known to return
   at most one element (query_one) or exactly one element (query_value).
   Corresponding execute_one() and execute_value() functions for prepared
   queries are also provided. For details, refer to Sections 4.3, "Executing
   a Query" and 4.5, "Prepared Queries" in the ODB manual.

 * Support for defining persistent objects as instantiations of C++ class
   templates, similar to composite value types. For details, refer to
   Section 15.2, "Persistent Class Template Instantiations" in the ODB
   manual.

 * Support for object and table join types in views. Supported join type
   are left, right, full, inner, and cross with left being the default.
   For details, refer to Sections 10.1, "Object Views" and 10.3, "Table
   Views" in the ODB manual.

 * Support for result modifiers in view query conditions. Currently
   supported result modifiers are 'distinct' (which is translated to
   SELECT DISTINCT) and 'for_update' (which is translated to FOR UPDATE or
   equivalent for database systems that support it). For details, refer to
   Section 10.5, "View Query Conditions" in the ODB manual.

 * Support for persisting std::deque containers.

 * New pragma, on_delete, allows the specification of an on-delete semantics
   (translated to the ON DELETE SQL clause) for an object pointer. For more
   information, refer to Section 14.4.15, "on_delete" in the ODB manual.

 * Besides odb::stderr_tracer there is now odb::stderr_full_tracer that
   traces statement preparations and deallocations in addition to their
   executions. This new implementation can be useful when you want to see
   the text of a statement that contains a syntax error and therefore
   will not actually be executed. For more information, refer to Section
   3.13, "Tracing SQL Statement Execution" in the ODB manual.

 * ODB can now compile headers that use #pragma once instead of include
   guards.

 * User-supplied prologues and epilogues are now generated outside the
   pre.hxx/post.hxx includes. This allows the use of precompiled headers
   with the generated files.

 * Support for calling MySQL stored procedures. For details and limitations
   refer to Section 17.7, "MySQL Stored Procedures" in the ODB manual.

 * Support for calling SQL Server stored procedures. For details and
   limitations refer to Section 21.7, "SQL Server Stored Procedures" in
   the ODB manual.

 * New option, --oracle-warn-truncation, makes ODB warn about SQL names
   that are longer than 30 characters and are therefore truncated. ODB
   now also detects when such truncations lead to Oracle name conflicts
   and issues diagnostics even without this option specified.

 * For MySQL, PostgreSQL, and SQL Server, ODB now warns when an SQL name
   exceeds the database limit (64, 63, and 128 characters, respectively).
   SQLite has no limitation on name lengths. For Oracle, which has a limit
   that is much more likely to be reached in normal circumstances (30
   characters), a more comprehensive detection is implemented (see the item
   above).

 * New option, --statement-regex, can be used to process prepared statement
   names that are used by PostgreSQL. This can be useful, for example, to
   shorten names that exceed the PostgreSQL name limit.

 * The --std option now accepts the 'c++14' value.

Version 2.3.0

 * Support for database schema evolution, including schema migration, data
   migration, and soft model changes. For more information, refer to Chapter
   13, "Database Schema Evolution" in the ODB manual.

 * Support for object sections. Sections are an optimization mechanism that
   allows the partitioning of data members of a persistent class into groups
   that can be loaded and/or updated separately. For more information, refer
   to Chapter 9, "Sections" in the ODB manual as well as the 'section'
   example in the odb-examples package.

 * Support for automatic mapping of C++11 enum classes in addition to "old"
   enums. For more information, refer to the ODB manual "Type Mapping"
   sections for each database system.

 * Support for defining composite value types inside persistent classes,
   views, and other composite values. For more information, refer to Section
   7.2, "Composite Value Types" in the ODB manual.

 * Support for pattern matching (SQL LIKE operator) in the C++-integrated
   queries. For more information, refer to Section 4.1, "ODB Query Language"
   in the ODB manual.

 * The schema_catalog::create_schema() function now has a third argument
   which indicates whether to drop the schema prior to creating the new one.
   The default is true which is backwards-compatible. The schema_catalog
   class now also provides the drop_schema() function which allows you to
   drop the schema without creating the new one. Finally, the exists()
   function now has the schema name argument which by default is an empty
   string (the default schema name). For more information, refer to Section
   3.4, "Database" in the ODB manual.

 * The transaction class now provides the default constructor that allows
   the creation of finalized transactions which can then be re-initialized
   with the reset() function. The finalized() accessor has also been added
   which allows querying of the finalization state. For more information,
   refer to Section 3.5, "Transactions" in the ODB manual.

 * New option, --fkeys-deferrable-mode, specifies the alternative deferrable
   mode for foreign keys. By default, the ODB compiler generates deferred
   foreign keys for databases that support them (SQLite, PostgreSQL, and
   Oracle) and comments the foreign keys out for databases that don't (MySQL
   and SQL Server). This option can be used to override this behavior. Refer
   to the ODB compiler command line interface documentation (man pages) for
   details.

 * Starting with MySQL version 5.6.4 it is possible to store fractional
   seconds up to microsecond precision in TIME, DATETIME, and TIMESTAMP
   columns. Both Boost and Qt profiles have been updated to support this
   new functionality. Note, however, that to enable sub-second precision,
   the corresponding type with the desired precision has to be specified
   explicitly. For details, refer to the "MySQL Database Type Mapping"
   sections in the Boost and Qt profile chapters.

 * New SQLite-specific exception, odb::sqlite::forced_rollback, which is
   thrown if SQLite forces a transaction to roll back. For more information,
   refer to Section 16.5.6, "Forced Rollback" in the ODB manual.

 * New options, --pgsql-server-version, can be used to specify the minimum
   PostgreSQL server version with which the generated C++ code and schema
   will be used. Right now this information is used to enable the use of
   the IF NOT EXISTS clause in the CREATE TABLE statement for the schema
   version table creation in PostgreSQL 9.1 and later. Refer to the ODB
   compiler command line interface documentation (man pages) for details.

 * The --output-name option has been renamed to --input-name, which is more
   semantically correct.

 * The generated database schema now explicitly specify NULL for nullable
   columns.

 * The generated separate C++ schema file (--schema-format separate) no
   longer includes the generated header file (-odb.hxx). As a result, it is
   now possible to use the --generate-schema-only and --at-once options to
   generate a combined C++ schema file for several headers.

Version 2.2.0

 * Multi-database support. This mechanism allows an application to
   simultaneously work with multiple database systems and comes in two
   flavors: static and dynamic. With static support the application uses
   the static database interfaces (that is, odb::<db>::database instead
   of odb::database). With dynamic support the same application code can
   access multiple databases via a common interface. Dynamic multi-database
   supports also allows the application to dynamically load the database
   support code for individual database systems if and when necessary. For
   more information, refer to Chapter 14, "Multi-Database Support" in the
   ODB manual.

 * Support for prepared queries. Prepared queries are a thin wrapper around
   the underlying database system's prepared statements functionality. They
   provide a way to perform potentially expensive query preparation tasks
   only once and then execute the query multiple times. For more information,
   refer to Section 4.5, "Prepared Queries" in the ODB manual as well as the
   'prepared' example in the odb-examples package.

 * Mapping for char[N] and std::array<char, N> to database VARCHAR(N-1) (or
   similar) as well as for char to database CHAR(1) (or similar). For SQL
   Server and SQLite on Windows equivalent mappings for wchar_t are also
   provided. Also the query support for arrays has been improved to allow
   passing a value of the decayed type (pointer) as a query parameter.
   For more information, refer to the ODB manual "Type Mapping" sections
   for each database system.

 * Support for change-tracking std::vector and QList container equivalents.
   Change-tracking containers minimize the number of database operations
   necessary to synchronize the container state with the database. For
   more information, refer to Sections 5.4, "Change-Tracking Containers",
   5.4.1 "Change-Tracking vector", and 22.3.1, "Change-Tracking QList"
   in the ODB manual.

 * Support for automatically-derived SQL name transformations (table, column,
   index, etc). At the higher level, it is possible to assign prefixes and
   suffixes (--table-prefix, --{index,fkey,sequence}--suffix options) as
   well as to convert to upper or lower case (--sql-name-case option). At
   the lower level, it is possible to specify transformations as regular
   expressions (--{table,column,index,fkey,sequence,sql-name}-regex options).
   For more information, refer to the SQL NAME TRANSFORMATIONS section in
   the ODB compiler command line interface documentation (man pages).

 * New options, --export-symbol and --extern-symbol, allow DLL-exporting of
   the generated database support code.

 * Support for transaction post- commit/rollback callbacks. For more
   information, refer to Section 13.1, "Transaction Callbacks" in the ODB
   manual.

 * Support for custom session implementations. For more information, refer
   to Section 10.2, "Custom Sessions" in the ODB manual.

 * Support for early connection release. Now the database connection is
   released when commit()/rollback() is called rather than when the
   transaction instance goes out of scope.

 * New odb::schema_catalog function, exists(), can be used to check whether
   a schema with the specified name exists in the catalog.

 * Support for SQL Server ROWVERSION-based optimistic concurrency. For more
   information, refer to Section 19.1.1, "ROWVERSION Support" in the ODB
   manual.

 * Support for specifying the SQL Server transaction isolation level. For
   more information, refer to Section 19.2, "SQL Server Database Class" in
   the ODB manual.

 * Support for "smart" containers. A smart container is provided with
   additional functions which allow it to insert, update, and delete
   individual elements in the database. Change-tracking containers are
   examples of smart containers that utilizes this new functionality.
   Currently only ordered smart containers are supported. Note also that
   with this addition the names of the database functions provided by the
   ODB compiler (see libodb/odb/container-traits.hxx) have changed. This
   will only affect you if you have added ODB persistence support for a
   custom container.

Version 2.1.0

 * The ODB compiler is now capable of automatically discovering accessor and
   modifier functions for inaccessible data members in persistent classes,
   composite value types, and views. It will then use these accessors and
   modifiers in the generated code instead of trying to access such data
   members directly. The names of these functions are derived from the
   data member names and, by default, the ODB compiler will look for names
   in the form: get_foo/set_foo, getFoo/setFoo, getfoo/setfoo, and just
   foo. You can also add custom name derivations with the --accessor-regex
   and --modifier-regex ODB compiler options. For more information, refer
   to Section 3.2, "Declaring Persistent Objects and Values" in the ODB
   manual.

 * New pragmas, get, set, and access, allow the specification of custom
   accessor and modifier expressions for data members in persistent classes,
   composite value types, and views. For more information, refer to Section
   12.4.5, "get/set/access" in the ODB manual as well as the 'access' example
   in the odb-examples package.

 * New pragma, virtual, allows the declaration of virtual data members. A
   virtual data member is an imaginary data member that is only used for
   the purpose of database persistence. This mechanism can be useful to
   aggregate or dis-aggregate real data members, implement the pimpl idiom,
   and to handle third-party types for which names of real data members may
   not be known. For more information, refer to Section 12.4.13, "virtual"
   in the ODB manual as well as the 'access' and 'pimpl' examples in the
   odb-examples package.

 * Support for defining database indexes. Both simple and composite indexes
   can be defined with support for database-specific index types, methods,
   and options. For more information, refer to Section 12.6, "Index
   Definition Pragmas" as well as Sections [13-17].16, "<Database> Index
   Definition" in the ODB manual.

 * Support for mapping extended database types, such as geospatial types,
   user-defined types, and collections. This mechanism allows you to map
   any database type to one of the types for which ODB provides built-in
   support (normally string or binary). The text or binary representation
   of the data can then be extracted into a C++ data type of your choice.
   For more information, refer to Section 12.7, "Database Type Mapping
   Pragmas" in the ODB manual.

 * The Boost profile now provides persistence support for the Boost Multi-
   Index container (boost::multi_index_container). For more information,
   refer to Section 19.3, "Multi-Index Container Library" in the ODB manual.

 * The Boost profile now provides persistence support for the Boost uuid type
   (boost::uuids::uuid). For more information, refer to Section 19.6, "Uuid
   Library" in the ODB manual as well as the 'boost' example in the odb-
   examples package.

 * The Qt profile now provides persistence support for the QUuid type. For
   more information, refer to Section 20.1, "Basic Types" in the ODB manual
   as well as the 'qt' example in the odb-examples package.

 * SQLite improvements: Persistence support for std::wstring on Windows
   (Section 14.1, "SQLite Type Mapping"). Ability to pass the database
   name as std::wstring on Windows (Section 14.2, "SQLite Database Class").
   Ability to specify the virtual filesystem (vfs) module in the database
   constructors (Section 14.2, "SQLite Database Class").

 * Support for mapping C++11 std::array<char, N> and std::array<unsigned
   char, N> types to BLOB/BINARY database types. For more information,
   refer to Sections [13-17].1, "<Database> Type Mapping" in the ODB manual.

 * Support for mapping the char[16] array to PostgreSQL UUID and SQL Server
   UNIQUEIDENTIFIER types. For more information, refer to Sections 15.1,
   "PostgreSQL Type Mapping" and 17.1, "SQL Server Type Mapping" in the
   ODB manual.

 * New option, --output-name, specifies the alternative base name used to
   construct the names of the generated files. Refer to the ODB compiler
   command line interface documentation (man pages) for details.

 * New option, --generate-schema-only, instructs the ODB compiler to
   generate the database schema only. Refer to the ODB compiler command
   line interface documentation (man pages) for details.

 * New option, --at-once, triggers the generation of code for all the input
   files as well as for all the files that they include at once. Refer to
   the ODB compiler command line interface documentation (man pages) for
   details.

 * New options, --sql-interlude and --sql-interlude-file, allow the insertion
   of custom SQL between the DROP and CREATE statements in the generated
   database schema file.

 * New options, --omit-drop and --omit-create, trigger the omission of DROP
   and CREATE statements, respectively, from the generated database schema.

 * New ODB manual Section, 6.3 "Circular Relationships", explains how to
   handle persistent classes with circular dependencies that are defined
   in separate headers.

 * The id() pragma that was used to declare a persistent class without an
   object id has been renamed to no_id.

 * New pragma, definition, allows the specification of an alternative code
   generation point for persistent classes, views, and composite value
   types. This mechanism is primarily useful for converting third-party
   types to ODB composite value types. For more information, refer to
   Section 12.3.7, "Definition" in the ODB manual.

 * The session constructor now accepts an optional bool argument (true by
   default) which indicates whether to make this session current for this
   thread. For more information, refer to Chapter 10, "Session" in the ODB
   manual.

 * Simplified Oracle automatically-assigned object id implementation that
   does not rely on triggers.

 * Support for mapping boost::posix_time::ptime and QDateTime to the DATE
   Oracle type. For more information, refer to Sections 19.4.4 (Boost) and
   20.4.4 (Qt) in the ODB manual.

 * Default SQLite mapping for float and double now allows NULL since SQLite
   treats NaN FLOAT values as NULL. For more information, refer to Section
   14.1, "SQLite Type Mapping" in the ODB manual.

Version 2.0.0

  * Support for C++11. The newly supported C++11 standard library components
    include:
      - std::unique_ptr as object pointer or value wrapper
      - odb::lazy_unique_ptr lazy counterpart
      - std::shared_ptr/weak_ptr as object pointer or value wrapper
      - odb::lazy_shared_ptr/lazy_weak_ptr lazy counterparts
      - support for array, forward_list, and unordered containers
      - connection factory can be passed to the database constructor as
        std::unique_ptr instead of std::auto_ptr

    The ODB compiler now recognizes the --std option. Valid values for this
    option are 'c++98' (default) and 'c++11'. In the runtime libraries the
    C++11 support is header-only which means that the same build of a runtime
    library can be used in both the C++98 and C++11 modes. On UNIX, the tests
    and examples can be compiled in the C++11 mode by passing the necessary
    options to turn the C++ compiler into this mode (e.g., -std=c++0x GCC
    option). On Windows, the tests and examples are always built in the C++11
    mode with VC++ 10 and later. The new 'c++11' example in the odb-examples
    package shows ODB support for some of the C++11 features.

  * Support for polymorphism. Now a persistent class hierarchy can be
    declared polymorphic which makes it possible to persist, load, update,
    erase, and query objects of derived classes using their base class
    interfaces. For more information, refer to Chapter 8, "Inheritance" in
    the ODB manual as well as the 'inheritance/polymorphism' example in the
    odb-examples package.

  * Support for composite object ids. Now a composite value type can be used
    to declare an object id member. For more information, refer to Section
    7.2.1, "Composite Object Ids" in the ODB manual as well as the 'composite'
    example in the odb-examples package.

  * Support for the NULL value semantics for composite values. For more
    information, refer to Section 7.3, "Pointers and NULL Value Semantics"
    in the ODB manual.

  * New schema format (--schema-format), 'separate', allows the generation
    of the schema creation code into a separate C++ source file (called
    '<name>-schema.cxx' by default). This value is primarily useful if you
    want to place the schema creation functionality into a separate program
    or library.

  * New namespace-level pragmas: table, pointer. The table pragma specifies
    the table prefix that is added to table names for all the persistent
    classes inside a namespace. The pointer pragma specifies the default
    pointer type to be used for persistent classes and views inside a
    namespace. For more information, refer to Section 12.5.1, "pointer" and
    Section 12.5.2, "table" in the ODB manual.

  * Session support is now optional and is disabled by default. This is a
    backwards-incompatible change. Session support can be enabled on the
    per class basis or at the namespace level using the new session pragma.
    It can also be enabled by default for all the persistent classes using
    the --generate-session ODB compiler option. Thus, to get the old behavior
    where all the objects were session-enabled, simply add --generate-session
    to your ODB compiler command line. For more information, refer to Chapter
    10, "Session" in the ODB manual.

  * The semantics of the database operations callbacks has changed with
    respect to object const-ness. This is a backwards-incompatible change.
    Now the callback function for the *_persist, *_update, and *_erase events
    is always called on the constant object reference while for the *_load
    events -- always on the unrestricted reference. For more information,
    refer to Section 12.1.7, "callback" in the ODB manual.

  * New function, transaction::reset(), allows the reuse of the same
    transaction instance to complete several database transactions. For more
    information, refer to Section 3.4, "Transactions" in the ODB manual.

  * New exception, odb::session_required, is thrown when ODB detects that
    correctly loading a bidirectional object relationship requires a session
    but one is not used. For more information, refer to Section 6.2,
    "Bidirectional Relationships" in the ODB manual.

Version 1.8.0

  * Support for the Microsoft SQL Server database. The provided connection
    factories include 'new' (a new connection is created every time one is
    requested) and 'pool' (a pool of connections is maintained). The Boost
    and Qt profiles have been updated to support this database. For more
    information, refer to Chapter 17, "Microsoft SQL Server Database" in
    the ODB manual.

  * Support for defining composite value types as C++ class template
    instantiations. For more information, refer to Section 7.2, "Composite
    Value Types" in the ODB manual as well as the 'composite' example in the
    odb-examples package.

  * Support for database schemas ("database namespaces"). A schema can be
    specified for a persistent class, for a C++ namespace (the schema then
    applies to all the persistent classes within this namespace), and for a
    file with the --schema ODB compiler option. For more information, refer
    to Section 12.1.8, "schema" in the ODB manual.

  * The --default-schema option has been renamed to --schema-name.

  * The default Oracle mapping for std::string has changed from VARCHAR2(4000)
    to VARCHAR2(512).

Version 1.7.0

  * Support for the Oracle database. The provided connection factories
    include 'new' (a new connection is created every time one is requested)
    and 'pool' (a pool of connections is maintained). The Boost and Qt
    profiles have been updated to support this database. For more information,
    refer to Chapter 16, "Oracle Database" in the ODB manual.

  * Support for optimistic concurrency. For more information refer to Chapter
    11, "Optimistic Concurrency" in the ODB manual as well as the 'optimistic'
    example in the odb-examples package.

  * Support for read-only objects, composite value types, and data members.
    The new readonly pragma can be used to declare one of these entities as
    read-only. Constant data members are automatically treated as read-only.
    For more information, refer to Section 12.1.4 "readonly (object)",
    Section 12.3.6 "readonly (composite value)", and Section 12.4.10
    "readonly (data member)" in the ODB manual.

  * Support for persistent classes without object identifiers. Such classes
    have to be explicitly declared as not having an object id and they have
    limited functionality. For more information, refer to Section 12.1.5
    "id" in the ODB manual.

  * Support for SQL statement execution tracing. For more information, refer
    to Section 3.12 "Tracing SQL Statement Execution" in the ODB manual.

  * Support for mapping char[N], unsigned char[N], and std::vector<unsigned
    char> to the BLOB (or equivalent) types. For more information, refer to
    Chapters 13 (for MySQL), 14 (for SQLite), 15 (for PostgreSQL), and 16
    (for Oracle) in the ODB manual.

  * Query result iterator now provides the id() function which allows one
    to get the object id without loading the object. For more information,
    refer to Section 4.4 "Query Result" in the ODB manual.

  * Support for microsecond precision in Boost and Qt date-time types
    mapping to PostgreSQL date-time data types. Additionally, Qt QDateTime
    values stored in a PostgreSQL database can now be earlier than the UNIX
    epoch.

Version 1.6.0

  * New concept, view, is a C++ class that embodies a light-weight, read-
    only projection of one or more persistent objects or database tables
    or the result of a native SQL query execution. Some of the common
    applications of views include loading a subset of data members from
    objects or columns from database tables, executing and handling
    results of arbitrary SQL queries, including aggregate queries, as
    well as joining multiple objects and/or database tables using object
    relationships or custom join conditions. For more information refer
    to Chapter 9, "Views" in the ODB manual as well as the 'view' example
    in the odb-examples package.

  * New function, database::erase_query(), allows the deletion of the
    database state of multiple objects matching certain criteria. It uses
    the same query expression as the database::query() function. For more
    information, refer to Section 3.10, "Deleting Persistent Objects" in
    the ODB manual.

  * Support for value wrappers. An ODB value wrapper is a class template
    that wraps a value type or a container. Common examples of wrappers
    are smart pointers, holders, and "optional value" containers such as
    boost::optional. A wrapper can be transparent or it can handle the
    NULL semantics. To allow the easy conversion of value types that do
    not support the NULL semantics into the ones that do, the odb::nullable
    class template has been added. ODB now also includes built-in support for
    std::auto_ptr and std::tr1::shared_ptr smart pointers as value wrappers
    as well as for boost::shared_ptr and QSharedPointer via the Boost and Qt
    profiles. Currently, the NULL semantics is only supported for simple
    values but smart pointers can still be used with composite values and
    containers. For more information, refer to Section 7.3, "NULL Value
    Semantics" in the ODB manual.

  * Support for the boost::optional container in the Boost profile. A data
    member of the boost::optional type is mapped to a column that can have
    a NULL value. For more information, refer to Section 15.3 "Optional
    Library" in the ODB manual.

  * Support for mapping std::vector<char> to the BLOB (or equivalent) types.
    For more information, refer to Chapters 11 (for MySQL), 12 (for SQLite)
    and 13 (for PostgreSQL) in the ODB manual.

  * New option, --table-prefix, allows the specification of a prefix that
    is added to table and index names. For more information, refer to the
    ODB compiler command line interface documentation (man pages).

  * New ODB runtime library interface, odb::connection, represents a
    connection to the database. The primary use case for a connection is to
    execute native statements outside of a transaction. For more information,
    refer to Section 3.5, "Connections" in the ODB manual.

  * Support for multiplexing several transactions on the same thread. For
    more information, refer to Section 3.4, "Transactions" in the ODB
    manual.

  * All the concrete connection classes now have a second constructor which
    allows the creation of a connection instance from an already established
    underlying connection handle. The connection_pool_factory and, in case of
    SQLite, single_connection_factory now have a virtual create() function
    that can be overridden to implement custom connection establishment and
    configuration.

  * The query expression syntax for object pointers and composite values has
    changed. Now, instead of using the scope resolution operator ('::'), the
    member access via a pointer operator (->) is used for object pointers and
    the member access operator (.) is used for composite values. Examples of
    old and new syntax for pointers, old: query<employee>::employer::name,
    new: query<employee>::employer->name. For composites values, old:
    query<employee>::name::first, new: query<employee>::name.first.

  * SQLite ODB runtime now enables foreign key constraints checking by
    default. While this should not affect correct applications, due to
    bugs in SQLite DDL foreign keys support, you may need to temporarily
    disable foreign key constraints checking when re-creating the database
    schema (the sign that you may need to do so is the "foreign key
    constraint failed" exception thrown by the commit() function after the
    call to schema_catalog::create_schema()). For more information, refer
    to Section 12.5.3, "Foreign Key Constraints" in the ODB manual.

  * Support for specifying the client character set for the MySQL database.
    For more information, refer to Section 11.2, "MySQL Database Class" in
    the ODB manual.

  * Object cache maintained by a session no longer distinguishes between
    const and non-const objects. Instead, const objects are treated as
    non-const by casting away constness. For more information on this new
    behavior, refer to Section 9.1, "Object Cache" in the ODB manual.

Version 1.5.0

  * Support for the PostgreSQL database. The provided connection factories
    include 'new' (a new connection is created every time one is requested)
    and 'pool' (a pool of connections is maintained). The Boost and Qt
    profiles have been updated to support this database. For more information,
    refer to Chapter 13, "PostgreSQL Database" in the ODB manual.

  * New handling of the NULL semantics. Now, instead of being specified as
    part of the SQL type with the type pragma, there are separate null and
    not_null pragmas. The not_null pragma was used to control the NULL
    semantics of object pointers. Now the two pragmas are used consistently
    for object pointers and simple values (and, in the future, they will work
    for composite values and containers). To control the NULL semantics of
    the container's element values, the value_null and value_not_null pragmas
    have been added, similar to the value_type, value_column, etc., pragmas.
    For more information about the new mechanism, refer to Sections 10.2.3,
    10.2.8, 10.3.4, and 10.3.13 in the ODB manual.

    This is a backwards-incompatible change. Existing use cases that will
    require manual changes are listed below.

    For pragmas that apply to simple value types and data members of
    such types:

    #pragma db type("TEXT NOT NULL") => #pragma db type("TEXT")
    #pragma db type("TEXT NULL")     => #pragma db type("TEXT") null
    #pragma db type("TEXT")          => #pragma db type("TEXT") null

    For pragmas that apply to containers of pointers and data members of
    such types:

    #pragma db not_null              => #pragma db value_not_null

  * New pragma, default, allows the specification of the database default
    value. For more information, refer to Section 10.3.5, "default" in the
    ODB manual.

  * New pragmas, options, id_options, index_options, key_options, and
    value_options, allow the specification of additional column definition
    options. For more information, refer to Section 10.3.6, "options" in
    the ODB manual.

  * Support for database operations callbacks. Now a persistent class can
    register a callback function that will be called before and after every
    database operation (such as persist, load, update, or erase) is performed
    on an object of this class. A database operations callback can be used to
    implement object-specific pre and post initializations, registrations,
    and cleanups. For more information and an example, refer to Section
    10.1.4, "callback" in the ODB manual.

  * New option, --include-regex, allows the modification of the #include
    directive paths generated by the ODB compiler. This is primarily useful
    when placing the generated code into subdirectories and the #include
    directives have to be adjusted accordingly. The --include-regex-trace
    option is useful for debugging the expressions specified with
    --include-regex.

Version 1.4.0

  * New profile, qt, provides persistence support for the Qt framework. This
    version covers the most commonly used basic types, date-time types, smart
    pointers, and containers. The qt profile implementation is provided by the
    libodb-qt library. For more information refer to Chapter 13, "Profiles
    Introduction" and Chapter 15, "Qt Profile" in the ODB manual as well as
    the 'qt' example in the odb-examples package.

  * Support for non-polymorphic object inheritance, including the new abstract
    pragma. For more information refer to Chapter 8, "Inheritance" in the ODB
    manual as well as the 'inheritance' example in the odb-examples package.

  * Automatic mapping of C++ enumerations to suitable database types. In
    database systems that support enumeration types (such as MySQL), a C++
    enum is mapped to such a type (for example, ENUM('red', 'green', 'blue')
    in MySQL). Otherwise, it is mapped to a suitable integer type. Refer to
    Part II, "Database Systems" in the ODB manual for more details on the
    provided mappings.

  * New pragma, id_type, allows the specification of the native database type
    that should be used for data members designated as object identifiers. In
    combination with the type pragma, id_type allows you to map a C++ type
    differently depending on whether it is used in an ordinary member or an
    object id.

  * New options, --odb-prologue-file and --odb-epilogue-file, allow the
    inclusion of file contents into the ODB compilation.

  * Default mapping of the size_t and std::size_t types to a 64-bit integer
    database type irrespective of the platform width. This can be overridden
    with the type pragma.

Version 1.3.0

  * Support for the SQLite database. The provided connection factories include
    'new' (a new connection is created every time one is requested), 'single'
    (single connection is shared among all the callers), and 'pool' (a pool
    of connections is maintained). In multi-threaded applications the runtime
    uses the SQLite shared cache and unlock notification features to aid
    concurrency. For more information, refer to Chapter 11, "SQLite Database"
    in the ODB manual.

  * Support for database-specific profiles. Now the ODB compiler first looks
    for the <profile>-<database>.options file and only if this file is not
    found, does it fall back to <profile>.options.

  * Support for the GCC 4.6 plugin interface changes.

Version 1.2.0

  * New profile, boost, provides persistence support for the Boost libraries.
    This version covers the most commonly used types from the smart_ptr,
    unordered, and date_time libraries. The boost profile implementation is
    provided by the libodb-boost library. For more information refer to
    Chapter 11, "Profiles Introduction" and Chapter 12, "Boost Profile" in
    the ODB manual as well as the 'boost' example in the odb-examples package.

  * Support for embedded database schemas. The new option, --schema-format,
    allows the selection of the schema format. The valid values for this
    option are 'sql' for a standalone SQL file and 'embedded' for a schema
    embedded into the generated C++ code. The new odb::schema_catalog class
    provides support for accessing embedded schemas from within the
    application. For details refer to Section 3.3, "Database" in the ODB
    manual as well as the 'schema/embedded' example in the odb-examples
    package.

  * New exceptions: odb::recoverable, odb::connection_lost, and odb::timeout.
    The odb::recoverable exception is a common base class for all recoverable
    ODB exceptions. The other two exceptions plus odb::deadlock now inherit
    from this base. Refer to Section 3.5, "Error Handling and Recovery" for
    details.

  * Support for connection validation (ping) in MySQL connection_pool_factory.
    This transparently deals with the MySQL server closing connections after
    a certain period of inactivity.

  * New namespace, odb::core, contains using-declarations for the core ODB
    constructs, such as the database, transaction, etc. The primary use of
    this namespace is in the using-directives:

    using namespace odb::core;

    The above form should be preferred over the old variant:

    using namespace odb;

    The new approach brings all the essential ODB names into the current
    namespace without any of the auxiliary objects such as traits, etc., which
    minimizes the likelihood of conflicts with other libraries.  Note that you
    should still use the odb namespace when qualifying individual names, for
    example, odb::database.

  * New option aliases: -q for --generate-query and -s for --generate-schema.

  * Support for the default options file. Now, if configured, the ODB compiler
    loads the default options file (by default ../etc/odb/default.options,
    relative to the ODB compiler binary). This file can be used for
    installation-wide customization, such as adding extra include search
    paths.

Version 1.1.0

  * Support for storing containers in the database. For more information refer
    to Chapter 5, "Containers" in the ODB manual as well as the 'container'
    example in the odb-examples package.

  * Support for unidirectional and bidirectional object relationships,
    including lazy loading. For more information refer to Chapter 6,
    "Relationships" in the ODB manual as well as the 'relationship' and
    'inverse' examples in the odb-examples package.

  * Support for composite value types. For more information refer to Chapter
    7, "Composite Value Types" in the ODB manual as well as the 'composite'
    example in the odb-examples package.

  * Support for sessions. A session is an application's unit of work that
    may encompass several database transactions. In this version of ODB a
    session is just an object cache. For more information refer to Chapter
    8, "Session" in the ODB manual.

  * Support for custom object pointers that allows you to use smart pointers
    to return, pass, and cache persistent objects. See Section 3.2, "Object
    Pointers" in the ODB manual for details.

  * Support for native SQL statement execution. See Section 3.9, "Executing
    Native SQL Statements" in the ODB manual for details.

  * New option, --profile/-p, instructs the ODB compiler to use the specified
    profile library. See the ODB compiler command line manual for details.

  * Support for literal names (template-id, derived type declarator such as
    pointers, etc) in data member declarations. Now, for example, you can use
    std::vector<std::string> directly instead of having to create a typedef
    alias for it.

  * Support for inheritance from transient base types for object types and
    composite value types.

  * New example, 'schema/custom', shows how to map persistent C++ classes to
    a custom database schema.

  * New options, --odb-prologue, --odb-epilogue, allow inclusion of extra code
    into the ODB compilation process. This can be useful for making additional
    traits specializations or ODB pragmas known to the ODB compiler.

  * Support for persistent classes without default constructors. For objects
    of such classes only the load() and find() database functions that
    populate an existing instance can be used. Similarly, only the load()
    query result iterator function which populates an existing instance can
    be used.

Version 1.0.0

  * Initial release.