@version $Id: release-notes.txt 365233 2005-12-21 22:51:31Z tomdz $

========================================================================
ObJectRelationalBridge -- Bridging Java Objects and Relational Databases
========================================================================

Apache ObJectRelationalBridge (OJB) is an Object/Relational mapping tool
that provides transparent transactional persistence for Java Objects
against relational databases. OJB provides ODMG and JDO interfaces.

More information is available at http://db.apache.org/ojb

Please refer to our issue tracker for more details on issues with id [OJB-xxx]
http://issues.apache.org/jira/browse/OJB


---------------------------------------------------------------------
Release 1.0.4
---------------------------------------------------------------------
NEW FEATURES:
* Support for embedded & network Derby (only distinct count queries don't work yet)
* Added ability to prepare the database for OJB's unit tests via DdlUtils (instead of Torque)
  Note that this is required when running the tests against Derby
* [OJB-10] - Configurable Proxy generation (including CGLIB based generation)
* [OJB-31] - Configurable JDBC driver-setting setFetchSize
* [OJB-14] - Setting custom JDBC driver tuning options trough jdbc-connection-descriptor
* [OJB-6]  - Support for stored procedures returning ResultSet.
* [OJB-41] - ODMG-api: Allow cascading delete using auto-delete setting in metadata
* [OJB-66] - Allow to specify a sequence start element for SequenceManagerNextValImpl.
  Fixed for Oracle, PostgreSQL, MaxDB/SapDB, DB2. Additionally sequence properties
  like 'increment by', 'cache', 'order',... are supported too. See 'sequence manger'
  guide and javadoc
* Add new ODMG extensions and configuration properties. All properties can be set
  at runtime as global property in class ImplementationExt and for a specific
  transaction in class TransactionExt:
  - 'ordering' Allow to dis-/enable OJB's object ordering on transaction commit
  - 'implicitLocking' Dis-/enable OJB's implicit locking
* Add new property 'TxCheck' in OJB.properties file. When enabled, on PB store and
  delete calls OJB checks for a running PB-tx, if not found an error is logged in order to
  avoid store/delete calls without a running PB-tx (while development).

NOTES:
* Internal OJB class configuration files are changed, don't forget to replace
  these files on upgrade from older version (by default the internal used
  classes are located in repository_internal.xml).
* Improved performance comparison tests ('ant perf-test' and 'ant performance').
  Now the tests more fair for registered API's
* In order to facilitate [OJB-10] the constructor signature of VirtualProxy has been changed.  

CHANGES:
* INTERNAL API CHANGES:
  Pluggable class interfaces changed to support and fix OJB-31, OJB-6, OJB-63:
  - StatementsForClassIF
  - StatementManagerIF
  - Platform
  - RowReader
  - SqlGenerator
* Configurable ProxyFactory is now available. Both JDK-based (default) and CGLIB
  are provided. CGLIB-based generation allows for proxies to be generated for all
  persistent classes, and not just ones that implement an interface.
* Change in behavior of property 'useAutoCommit="1"' in jdbc-connection-descriptor
  detailed info can be found in reference guide of the repository file
* ConnectionFactoryManagedImpl is declared deprecated. Now OJB automatic detect the
  running JTA-transaction and suppress critical method calls on the used connection
* Removed deprecated SequenceManager implementation SequenceManagerMySQLImpl
* Remove unused method #setReferenceFKs from SequenceManager interface
* ODMG-api: Method behavior of ImplementationImpl#setImplicitLocking(boolean) changed.
  Now this method set the global property for 'implicitLocking'. Old version only changed the
  implicit locking behavior for the current used transaction, this is moved to
  TransactionExt#setImplicitLocking
* Internal used OJB_HL_SEQ table changed. Remove of deprecated column.
  If the SequenceManagerHighLowImpl was used and problems occur after upgrade, please
  drop and recreate this table without removed column
* Cleanup naming of 'PersistentField' implementation classes: Remove deprecated classes, rename
  implementation classes, declare all ...ImplNew classes as deprecated.
  See OJB.properties file.
* Upgraded ANTLR to v2.7.5 (OQL and JDOQL parser classes have been re-generated)
* Upgraded commons-dbcp to v1.2.1
* Upgraded commons-pool to v1.2
* Upgraded commons-beanutils to v1.7
* Upgraded commons-logging to v1.0.4
* Upgraded commons-collections to v3.1
* Upgraded hsql to v1.8.0.2
* Upgraded Torque to v3.1.1

BUG FIXES:
* [OJB-18] - ODMG ordering problem with circular/bidirectional 1:1 references
* [OJB-25] - Inheritance (each subclass on separate table) PK problem with
  nextval SequenceManager
* [OJB-26] - JDBC-ODBC bridge problem
* [OJB-27] - [ODMG] Foreign key violation
* [OJB-29] - Infinite loop in case of refresh=true and cycles in references
* [OJB-40] - CallableStatement check doesn't work with maxDB
* [OJB-42] - MsSQLServer, auto-generated querry error (m:n relation)
* [OJB-44] - ReferenceMap$Entry keeps growing over period of application usage.
* [OJB-52] - KeyConstraintViolatedException during store
* [OJB-54] - ConnectionFactoryManagedImpl may try to use pool
* [OJB-55] - Getting foreignkeyField's FieldDescriptor is failing if the foreignkey
  field is defined in the base class
* [OJB-59] - Incorrect identity management in SequenceManagerNativeImpl
  because of wrong plate-forme settings
* [OJB-62] - addPathClass in QueryByCriteria does not consider inheritance
* [OJB-63] - Bug when query against objects mapped on multiple joined tables ("super"-references)
* [OJB-70] - Connection reset by peer: socket write error
* [OJB-76] - OJB MtoN implementor failes to work properly with composite M&N keys that
  have some fields common.
* [OJB-77] - PlatformMsSQLServerImpl.getLastInsertIdentityQuery creates incorrect SQL
  statement to pull the identity from the last

KNOWN ISSUES:
* Auto-Detection for insert/update objects checks to avoid DB queries the PK fields of the object.
  If at least one PK field is 'NULL' or if primitive field '0', OJB assume the object is new and
  needs insert. This will be configurable in next upcoming version. Workaround for PB-api: use method
  PB#store(Object obj, ObjectModification mod) to state update or insert.
* Batch handling doesn't work properly with optimistic locking. This will be fixed
  in version 1.1
* Subqueries are not extent aware. see QueryTest#testSubQueryAgainstExtents
* Managed Environment: When run OJB in managed environment and using PBStateListener, the call of
  PBStateListener#beforeClose(PBStateEvent event) is made twice when
  a PB instance was closed in a JTA-tx (for the first time when PB.close() was called
  in bean on the PB handle, second time when the JTA-tx completes)
* otm-api: The OTM API has a known caching issue and should not be considered for code used
  in production environments. The future of the OTM layer will be subject for discussion on the
  OJB developers list, if you are using it - please subscribe to ojb-dev and make your voice heard.
* Oracle9i platform: when using statement batching, there is a 2k limit on BLOB + 4k limit on CLOB.



---------------------------------------------------------------------
Release 1.0.3
---------------------------------------------------------------------
NEW FEATURES:


NOTES:
- odmg-api: Named objects can now be non- and persistence capable objects. At least
  the named object have to be serializable.

CHANGES:
- Refactoring DCollection implementations, remove DListImpl_2 classes (code moved to DListImpl)
- Object2ByteArrFieldConversion: If source object on conversion javaToSQL is 'null', no longer create an
  byte array simply return 'null'

BUG FIXES:
Please refer to our Bug tracking site (http://issues.apache.org/jira/secure/BrowseProject.jspa?id=10700)
to see details for a bug with id OJB-xxx.

- odmg-api: OJB-7 - Fix critical property setting in OJB.properties file
- odmg-api: OJB-8 - Fix bug when using cascading delete on empty 1:1 reference
- odmg-api: Fix bug with named object. On 'unbind' of a named object, if the object is a persistence capable
  object (object with metadata mapping), the object itself will not be deleted. All other (serializable)
  named objects will be deleted on unbind.
- odmg-api: Fix bug with named objects when using database based Identity columns (SequenceManagerNativeImpl)
- odmg-api: Fix bug when persisting DCollection instances when using database based
  Identity columns (SequenceManagerNativeImpl).



KNOWN ISSUES:
- Auto-Detection for insert/update objects checks to avoid DB queries the PK fields of the object.
  If at least one PK field is 'NULL' or if primitive field '0', OJB assume the object is new and
  needs insert. This will be configurable in next upcoming version. Workaround for PB-api: use method
  PB#store(Object obj, ObjectModification mod) to state update or insert.
- Batch handling doesn't work proper with optimistic locking. This will be fixed
  in version 1.1
- Subqueries are not extent aware. see QueryTest#testSubQueryAgainstExtents
- When a class mapped on multiple joined tables was used in a reference only objects
  of the base type class will be instantiated, e.g. a Company class has
  a 1:n reference 'employees' to a base class Employee and class Manager extends Employee,
  then 'employees' only contains objects of type Employee even if the real type was Manager.
  See in OJB test suite ...broker.InheritanceMultipleTableTest#testInheritancedObjectsInCollectionReferences
  Same problem occur when query the base class, then OJB only returns objects of base type instead
  the real type.
- Managed Environment: When run OJB in managed environment and using PBStateListener, the call of
  PBStateListener#beforeClose(PBStateEvent event) is made twice when
  a PB instance was closed in a JTA-tx (for the first time when PB.close() was called
  in bean on the PB handle, second time when the JTA-tx completes)
- otm-api: The OTM API has a known caching issue and should not be considered for code used
  in production environments. The future of the OTM layer will be subject for discussion on the
  OJB developers list, if you are using it - please subscribe to ojb-dev and make your voice heard.
- Oracle9i platform: when using statement batching, there is a 2k limit on BLOB + 4k limit on CLOB.

---------------------------------------------------------------------
Release 1.0.2
---------------------------------------------------------------------
NEW FEATURES:

- New ant-target ojb-quickstart creates archives of ready-to-run apps for tutorials 1 and 2
  complete with pre-created Hsqldb databases (based on ojb-blank).
  These archives are also available in the binary download area.
- Introduce first version of a two-level cache (transactional session cache with an application cache
  handle with copies of cached objects). More detailed information see 'object cache' reference guide in docs.
- All Oracle platforms will now transparently handle java.lang.Character fields mapped to
  jdbc-types CHAR or VARCHAR, without any conversion class.
- Oracle9i plaform now handles CLOB>4k and BLOB>2k when DBCP and/or P6Spy are used.
- Oracle9i plaform can now be used with Oracle10g JDBC-driver. Escape processing statements are
  removed and OracleConnection unwrapping have been adjusted to be compatible with 10g JDBC.
- XDoclet module:
    * Important: generate-table-info now only prevents the generation of a table in the database schema,
      the class-descriptor is unaffacted
    * The new ojb.class#generate-repository-info attribute prevents the generation of field/reference/collection
      descriptors in the repository as well as the table in the database schema
      This attribute should be used when using inheritance with interfaces/abstract classes
    * Database foreignkeys are now generated for collections, as well
    * It is possible to prevent the generation of database foreignkeys for individual references
      or collections using the database-foreignkey attribute (ojb.collection/ojb.reference).
      However this attribute cannot be used to force generation of database-foreignkeys
    * Inheritance with interfaces/abstract classes (i.e. classes without table) is now handled properly
      for types that are referenced by references/collections even if the relevant foreignkey
      (reference/1:n collection) or primarykey (m:n collection) is not defined in the basetype itself
      but only in the subtypes. The XDoclet module will determine these fields and generate virtual
      field descriptors in the class-descriptor of the basetype (even if generate-repository-info=false).
      Additionally, database foreignkeys are generated if the subtypes map to the same table
    * Additional attributes for generating documentation in the database schema:
        - ojb.class#table-documentation
        - ojb.field#column-documentation
        - ojb.collection#indirection-table-documentation
        -               #foreignkey-documentation
        -               #remote-foreignkey-documentation
    * Foreignkey columns of an m:n collection can be made primarykeys of the indirection table by
      using the ojb.collection#indirection-table-primarykeys attribute

NOTES:
- ODMG-api refactoring!! All known issues in test suite are fixed. But the refactored odmg-api
  version needs changes in auto-xxx settings in metadata and changed OJB.properties settings.
  Please read carefully the 'CHANGES notes' below.
- Don't forget to replace the old repository.dtd and OJB.properties file (both files changed).
- If you plan to use OJB with an J2SE version older than 1.4, then you have to
  replace the Geronimo jars with the corresponding ones from Sun's J2EE SDK. This
  is because Geronimo is a J2EE 1.4 implementation and thus dependant upon
  J2SE 1.4. Use instead the version of Sun's J2EE reference implementation that
  correspond to your J2SE version. E.g. for J2SE 1.3 use the J2EE 1.3 SDK which
  you can get from here: http://java.sun.com/j2ee/1.3/index.jsp
- Performance improvement in handling of m:n relations
- Add new property 'sequenceStart' for SequenceManagerHighLowImpl and SequeceManagerInMemoryImpl
  more info see docs section 'sequence manager'
- Recommended to read the updated 'object cache' reference guide to reflect on the changes made in
  caching behavior.
- By default a new locking implementation was used for odmg, to enable the old deprecated odmg-locking,
  please see comments in OJB.properties file.

CHANGES:
- When using Oracl9i platform and DBCP connection factory, no PreparedStatement caching will occur
  in the DBCP pool (Oracle JDBC-driver statement caching is performed by the platform as default).
  statement pool, to avoid running out of server-side cursors when using Oracle9i platform.
- Oracle9i default platform should now be usable and have large LOB support in JBoss environemnts,
  please try the default Oracle9i platform and report your results to the ojb-user list.
- Oracle9i default platform should now be usable and have large LOB support in BEA WebLogic,
  please try the default Oracle9i platform in favor of WLOracle9i and report your results to
  the ojb-user list.
- Oracle9i platform no longer throws exceptions when Oracle-specific extensions initializer fails,
  this should enable Oracle9i platform to degrade smoothly in not yet handled managed environments.
- Oracle9i platform now uses ANSI SQL92 join syntax as per user-request. The Oracle platform
  still uses Oracle join syntax to be compatible with Oracle 8 and older versions.
- CollectionProxy classes will now throw an exception in constructor, if trying
  to use dynamic proxies with MetadataManager in 'per thread changes' mode but
  without any metadata profile key loaded in the current thread. See note about
  fix in CollectionProxy under "BUG FIXES" below.
- Minor changes in the repository.dtd. Add element object-cache to metadata interface and abstract class
  declaration in class-descriptor. So replacement of old repository.dtd is needed.
- Remove object-cache declaration from OJB.properties file. Use the 'object-cache' element in
  repository file (example see repository_database.xml file shipped with OJB) to specify the cache
  implementation. More detailed information see caching reference guide in docs.
- Remove redundant ObjectCache implementations (ObjectCacheSoftImpl, ObjectCacheUnlimitiedImpl), same
  behavior is possible with ObjectCacheDefaultImpl.
- Remove undocumented 'ObjectCacheFilter' (used to filter out whole packages or classes from being cached).
  Filter out packages or classes from being cached is still possible, please see 'object cache' reference guide.
- !!!In managed enviroments the org.odmg.Transaction#abort() call no longer throws an
  TransactionAbortedExceptionOJB, instead OJB does internal cleanup and set used
  JTA-tx to setRollbackOnly if possible. Thus the client does no longer get an RemoteException.
  The user have to throw an EJBException by himself to notify the client.
- Performance of the default RowReader (RowReaderDefaultImpl) was improved, when mapping
  several classes to the same table ("ojb-conrete class" feature), now only the fields of the
  associated class were performed in RowReader instead invoking all fields of the table used by
  the mapped classes.
- Introduce new locking-package in kernel api. Now top-level api like odmg-api can use a kernel
  lock manager to provide locking. Adapt odmg-locking to the new lock management and declare old locking stuff
  in org.apache.ojb.odmg.locking package as deprecated. Add support for apache's commons-transaction locking part.
- Use of database identity column (SequenceManagerNativeImpl). Move assign of PK values from
  PersistenceBrokerImpl to JdbcAccessImpl#executeInsert. Remove usage of SequenceManager#setReferenceFKs,
  will be handled by OJB in same way as without usage of database identity column.
- odmg-api: Introduced new object reordering implementation (replaces old algorithm
  in ObjectEnvelopeTable).
- odmg-api: Change used 'OqlCollectionClass' in OJB.properties from a DList impl to ArrayList impl
  to improve performance. If you need the additional features of odmg DList comment in the DList impl.
- odmg-api: ** All relations (1:1, 1:n and m:n) need auto-update/delete setting 'none' to proper work.**
- odmg-api: Cascade delete is now configurable via OJB.properties file and at runtime using
  TransactionExt#setCascadingDelete method (cast Transaction instance to TransactionExt)

BUG FIXES:
Please refer to our Bug tracking site (http://issues.apache.org/scarab/servlet/scarab/)
under http://issues.apache.org/scarab/servlet/scarab/issues/id/OJBxxx
to see details for a bug with id OJBxxx.

- Fix problem with StatementManager when binding delete-statement parameters
  with null-values for numeric type columns under certain conditions.
  (Oracle9i and jdbc-type=NUMERIC with a conversion class that can return null-values
   was such a combination, resulting in SQLException in OracleStatement#get_internal_type.)
- Fix bug in MetadataManager, when 'per thread changes' is enabled and CollectionProxy
  proxies are used. Using proxies in this mode used to be undefined, since the lazy
  loading might or might not succeed depending on in which thread the data was loaded.
  OJB will now reload the metadata profile used when creating the proxy,
  before attempting to reference any persistence capable classes.
  (Fixes occasional ClassNotPersistenceCapable exceptions seen in multithreaded
   applications when using 'per thread changes'.)
  See http://mail-archives.apache.org/eyebrowse/ReadMsg?listName=ojb-dev@db.apache.org&msgNo=9143
- Managed Environment: Fix problem of abandoned connections and statements (e.g. detected
  by JBoss) when query (iterator) result sets only partially materialized.
- Fix bug in ObjectCacheDefaultImpl resolve cache object conflicts, see
  http://mail-archives.apache.org/eyebrowse/ReadMsg?listName=ojb-user@db.apache.org&msgNo=15063
- LockManager#releaseLock, when a write lock was found all read locks will be released too.
  http://mail-archives.apache.org/eyebrowse/ReadMsg?listName=ojb-user@db.apache.org&msgNo=14885
- Fix problem with automatic generated sequence names in SequenceManagerHighLowImpl and
  SequenceManagerInMemoryImpl when different DB use the same table and column name for auto
  incremented fields.
  See http://mail-archives.apache.org/eyebrowse/ReadMsg?listName=ojb-user@db.apache.org&msgNo=14528
- Fix materialization problem under heavy load, OJB no longer returns partially materialized
  objects from cache.
  See http://mail-archives.apache.org/eyebrowse/ReadMsg?listId=107&msgNo=14472
- Fix key constraint problems when insert m:n relation manually via "linking"
- Fix bug in SequenceManagerNativeImpl used to support DB identity columns, make counter for
  temporary keys static to prevent lock not granted exceptions on insert of new objects.
- odmg-api: if within a transaction the client returns a different instance of an already
  locked object (e.g. serialized through network) method tx.lock and tx.markDirty do not
  replace the already locked object instance by the new one. Now it will do so.
- odmg-api: In managed environment odmg-api does clenup internal stuff (cache, locking) when
  Synchronization#afterCompletion(int status) was called by the JTA-TxManager instead doing all this
  stuff in #beforeCompletion()
- odmg-api: Assign FK for 1:1 relations when persistent objects are written to DB, instead when objects
  were locked.
- odmg-api: Mixing of pessimistic and optimistic locking should be possible. Now all lock calls on objects
  with optimistic locking enabled will be ignored.
- odmg-api: Fixed: If a user exchange already existing objects in 1:n references without changing the size
  of the collection, the main object will not become dirty and the FK values of the exchanged objects
  will not be updated. This is fixed.
- odmg-api: Fixed: Creation of m:n relation only works when objects created step by step (or use PB-api
  as workaround), persist a whole object graph seems not to work proper. This is fixed.
- odmg-api: Fixed: Mapping classes on multiple joined tables was not supported in
  ODMG-API implementation. This is fixed (!except the known issue when Query or refer to a base class).

KNOWN ISSUES:
- Auto-Detection for insert/update objects checks to avoid DB queries the PK fields of the object.
  If at least one PK field is 'NULL' or if primitive field '0', OJB assume the object is new and
  needs insert. This will be configurable in next upcoming version. Workaround for PB-api: use method
  PB#store(Object obj, ObjectModification mod) to state update or insert.
- Batch handling doesn't work proper with optimistic locking. This will be fixed
  in version 1.1
- Subqueries are not extent aware. see QueryTest#testSubQueryAgainstExtents
- When a class mapped on multiple joined tables was used in a reference only objects
  of the base type class will be instantiated, e.g. a Company class has
  a 1:n reference 'employees' to a base class Employee and class Manager extends Employee,
  then 'employees' only contains objects of type Employee even if the real type was Manager.
  See in OJB test suite ...broker.InheritanceMultipleTableTest#testInheritancedObjectsInCollectionReferences
  Same problem occur when query the base class, then OJB only returns objects of base type instead
  the real type.
- Managed Environment: When run OJB in managed environment and using PBStateListener, the call of
  PBStateListener#beforeClose(PBStateEvent event) is made twice when
  a PB instance was closed in a JTA-tx (for the first time when PB.close() was called
  in bean on the PB handle, second time when the JTA-tx completes)
- otm-api: The OTM API has a known caching issue and should not be considered for code used
  in production environments. The future of the OTM layer will be subject for discussion on the
  OJB developers list, if you are using it - please subscribe to ojb-dev and make your voice heard.
- Oracle9i platform: when using statement batching, there is a 2k limit on BLOB + 4k limit on CLOB.


---------------------------------------------------------------------
Release 1.0.1
---------------------------------------------------------------------
NEW FEATURES:
- Add new service for Identity object creation - PersistenceBroker#serviveIdentity()
  This service makes creation of Identiy objects to lookup objects by primary key much
  easier. More info see PersistenceBroker Tutorial section "Find object by primary key".
    For example, to lookup an object by its Identity with PB.getObjectByIdentity, for
    a single Long PK field (with value '23') you can call:
    Identity oid = broker.serviceIdentity().buildIdentity(MyClass.class, new Long(23));
    result = (MyClass) broker.getObjectByIdentity(oid);

NOTES:
- Changed build dependency from Sun's SCSL licensed J2EE specification jars to the
  Geronimo projects's ASL 2.0 licensed J2EE specification jars. Users now can build
  the core of OJB (everything but the jdori plugin) directly from the source tarball
  or CVS.
  NOTE: If you plan to use OJB with an J2SE version older than 1.4, then you have to
        replace the Geronimo jars with the corresponding ones from Sun's J2EE SDK. This
        is because Geronimo is a J2EE 1.4 implementation and thus dependant upon
        J2SE 1.4. Use instead the version of Sun's J2EE reference implementation that
        correspond to your J2SE version. E.g. for J2SE 1.3 use the J2EE 1.3 SDK which
        you can get from here: http://java.sun.com/j2ee/1.3/index.jsp

CHANGES:
- PersistenceBroker interface changed, add service method for IdentityFactory
- Performance improvement in RsIterator, first only map PK values from the
  result set and check against the cache. If no match map all field values
  from result set to internal Map class (old version always maps all values
  from the result set to the internal Map class).
- add in ObjectCacheDefaultImpl new attribute 'cachingKeyType' this allows to
  differ cached objects of same class for different databases.
  http://mail-archives.apache.org/eyebrowse/ReadMsg?listName=ojb-user@db.apache.org&msgNo=13931

BUG FIXES:
Please refer to our Bug tracking site (http://issues.apache.org/scarab/servlet/scarab/)
under http://issues.apache.org/scarab/servlet/scarab/issues/id/OJBxxx
to see details for a bug with id OJBxxx.

- fixed issue in AnonymousPersistentFieldForInheritance, class now use
  specified factory methods (in class-descriptor) for pc objects too
- fix for OJB275, include PK fields for "super" class-descriptor when using
  "super" reference. Now on object insert, the PK values are copied from the
  main object to the temporary created "super" objects.
- fixed issue in PersistenceBrokerFactorySyncImpl, always wrap requested PB
  instance with PB handle object
- fixed issue in ObjectCachePerClassImpl, allow to set 'class-cache' to null
- fixed issue with CollectionProxy and RemovalAwareCollection
- RemovalAwareCollection is no longer the default collection-class for m:n-relationships
- ManageableCollection#afterStore is only called if cascadeStoring is CASCADE_OBJECT
- ojb no longer adds groupBy columns to the SELECT-clause, orderBy columns are still added because of sapdb
- orderBy columns of the original query are used as joinAttributes in count-query
- ALIAS-prefix again works for report-queries (see QueryTest#testReportQueryAlias)
- odmg-api: Fix problem with TransactionExt#flush() call, always check for
  changed objects
- odmg-api: Fix issue with FK assignment, referenced proxy object should not be materialized to assign
  FK in main object.
- odmg-api: Fix Database#open(...) issue. In managed environments now no running JTA-tx was expected  

KNOWN ISSUES:
- odmg-api: If a user exchange already existing objects in 1:n references without changing the size
  of the collection, the main object will not become dirty and the FK values of the exchanged objects
  will not be updated.
  E.g. two objects obj_1 anf obj_2 with 1:n reference to ref objects, each with one
  existing/persistent reference object, obj_1{ref_1} and obj_2{ref_2}.
  Lock objects and exchange the references in collection obj_1{ref_2}
  and obj_2{ref_1} and commit --> FK values of ref_1 and ref_2 will not be updated.
- odmg-api: Creation of m:n relation only works when objects created step by step (or use PB-api
  as workaround), persist a whole object graph seems not to work proper.
- Batch handling doesn't work proper with optimistic locking. This will be fixed
  in version 1.1
- Subqueries are not extent aware. see QueryTest#testSubQueryAgainstExtents
- When using native Identity columns, OJB uses a temporary dummy value for created OJB Identity objects 
  of new pc objects (negative long values are used as dummy values for Identity columns, the real value 
  is available after the object insert), so the FK assignment is only valid after store of the 
  referenced object (PB-api handles this correctly). TransactionImpl#lock assign the FK before the 
  referenced object was written to DB. This only effects the ODMG API.
- Mapping of an inheritance hierarchy to multiple joined tables fails in the ODMG
  API only.

---------------------------------------------------------------------
Release 1.0
---------------------------------------------------------------------
NEW FEATURES:
--

NOTES:
--

CHANGES:
- We introduce new PersistentField implementations in OJB.properties file called
PersistentFieldXXXXImplNew. Per default one of the new PersistentField implementations was
used. If you hit problems please send a report. You can reverting to old impl by
changing property 'PersistentFieldClass' in OJB.prperties file. The new versions
has much better performance in read/write nested fields (up to 50 times faster than
the 'old' implementions) and will replace the old classes on next version. Because
PersistentField is an important kernel class we wait for community response before
final replacement.

BUG FIXES:
Please refer to our Bug tracking site (http://issues.apache.org/scarab/servlet/scarab/)
under http://issues.apache.org/scarab/servlet/scarab/issues/id/OJBxxx
to see details for a bug with id OJBxxx.


KNOWN ISSUES:
- The default ojb escape character '\' does not work for database using the same default as well and
should therefore be double-escaped. As a workaround always use a custom escape character:

	LikeCriteria.setEscapeCharacter('|');
	Criteria crit = new Criteria();
	crit.addLike("firstname", "h%|%");

This issue is fixed for MySql and PostgreSql so far.

- odmg-api: If a user exchange already existing objects in 1:n references without changing the size
of the collection, the main object will not become dirty and the FK values of the exchanged objects
will not be updated.
E.g. two objects obj_1 anf obj_2 with 1:n reference to ref objects, each with one
existing/persistent reference object, obj_1{ref_1} and obj_2{ref_2}.
Lock objects and exchange the references in collection obj_1{ref_2}
and obj_2{ref_1} and commit --> FK values of ref_1 and ref_2 will not be updated.

- odmg-api: Creation of m:n relation only works when objects created step by step (or use PB-api
as workaround), persist a whole object graph seems not to work proper.

- ReportQueries should not be used with columns referencing Classes with extents:

    ReportQueryByCriteria q = QueryFactory.newReportQuery(ProductGroup.class, crit);
    q.setAttributes(new String[] { "groupName", "sum(allArticlesInGroup.stock)", "sum(allArticlesInGroup.price)" });
    q.addGroupBy("groupName");

    ProductGroup.allArticlesInGroup points to class Article having multiple extents.

    As a workaround the query can be 'reversed' :

    ReportQueryByCriteria q = QueryFactory.newReportQuery(Article.class, crit);
    q.setAttributes(new String[] { "productGroup.groupName", "sum(stock)", "sum(price)" });
    q.addGroupBy("productGroup.groupName");

    Due to the fact the Article has extents multiple selects will be executed,
    so the same ProductGroup may show up more than once.

    Please see the testcases QueryTest#testReportQueryGroupByExtents1 and
    QueryTest#testReportQueryGroupByExtents2.

- A count on ReportQueries containing groupBy does not deliver a correct result.

- Batch handling doesn't work proper with optimistic locking. This will be fixed
before long after 1.0



---------------------------------------------------------------------
Release 1.0 rc7
---------------------------------------------------------------------

NEW FEATURES:

NOTES:
- OJB.properties file has changed, don't forget to replace on update!
 Among other things the property 'ImplementationClass' was needed to set in managed environments.

CHANGES:

-OJB.properties file new properties:
ListProxyClass
IndirectionHandlerClass
SetProxyClass
RowReaderDefaultClass
ImplementationClass
The logging properties moved to a separate file - see below.

- Logging settings have moved to separate OJB-logging.properties file

- Logging initialization is now decoupled form OJB initialization; this is
described in the new reference guide for logging

- It is no longer necessary to provide an empty repository.xml file when starting
without repository and connection descriptors

- rename/move internal package org.apache.ojb.odmg.transaction to org.apache.ojb.broker.transaction.tm
In managed environments each (top-level) API use transaction manager access, thus the TM related
classes are moved to the PB kernel and OJB.properties entries change.

- Base class for ODMG api access within non- or managed environments is now
org.apache.ojb.odmg.OJB. The used org.odmg.Implementation interface implementation
is specified in OJB.properties.

- ConnectionManager is more strict on CM.releaseConnection() method calls. Now an
exception is thrown when CM is in a "local transaction" status when try to release
the connection. The local tx status of ConnectionManager and PersistenceBroker implementation
is now decoupled, useful in managed environments allows to "close the connection" without
change the PB tx-state.

- the indirection handler (for reference proxies), and the list and set proxy classes
can now be configured in the OJB.properties file

- new CollectionProxy interface introduced to allow the ODMG api to make use of alternate collection proxy implementations.

BUG FIXES:

Please refer to our Bug tracking site (http://issues.apache.org/scarab/servlet/scarab/)
under http://issues.apache.org/scarab/servlet/scarab/issues/id/OJBxxx
to see details for a bug with id OJBxxx.

KNOWN ISSUES:
- odmg-api: If a user exchange already existing objects in 1:n references without changing the size
of the collection, the main object will not become dirty and the FK values of the exchanged objects
will not be updated.
E.g. two objects obj_1 anf obj_2 with 1:n reference to ref objects, each with one
existing/persistent reference object, obj_1{ref_1} and obj_2{ref_2}.
Lock objects and exchange the references in collection obj_1{ref_2}
and obj_2{ref_1} and commit --> FK values of ref_1 and ref_2 will not be updated.

- odmg-api: Creation of m:n relation only works when objects created step by step (or use PB-api
as workaround), persist a whole object graph seems not to work proper.

- ReportQueries should not be used with columns referencing Classes with extents:

    ReportQueryByCriteria q = QueryFactory.newReportQuery(ProductGroup.class, crit);
    q.setAttributes(new String[] { "groupName", "sum(allArticlesInGroup.stock)", "sum(allArticlesInGroup.price)" });
    q.addGroupBy("groupName");

    ProductGroup.allArticlesInGroup points to class Article having multiple extents.

    As a workaround the query can be 'reversed' :

    ReportQueryByCriteria q = QueryFactory.newReportQuery(Article.class, crit);
    q.setAttributes(new String[] { "productGroup.groupName", "sum(stock)", "sum(price)" });
    q.addGroupBy("productGroup.groupName");

    Due to the fact the Article has extents multiple selects will be executed,
    so the same ProductGroup may show up more than once.

    Please see the testcases QueryTest#testReportQueryGroupByExtents1 and
    QueryTest#testReportQueryGroupByExtents2.

- A count on ReportQueries containing groupBy does not deliver a correct result.

- Batch handling doesn't work proper with optimistic locking. This will be fixed
before long after 1.0

---------------------------------------------------------------------
Release 1.0 rc6
---------------------------------------------------------------------

NEW FEATURES:
 -

NOTES:
- Repository.dtd has changed, don't forget to replace this file
- Upgraded to the new Apache License, Version 2.0
- No support for HSQLDB 1.7.2RC1 yet
- No support for Torque 3.1 yet

- Add new PersistenceBrokerFactory implementation for use in managed environments when *only* the
PB-api was used and participation in JTA transaction via Synchronization interface is needed.

- Add new configuration property 'autoSync' to ObjectCacheDefaultImpl. Used to enable a simple
synchronization mechanism to keep cache in sync with DB

- Add new attribute values for reference-/collection-descriptor auto-update/-delete attribute.
Now we have five possible values: none, link, object, true, false. The true, false values
are now deprecated but will NOT be removed in near future (so don't panic!).

CHANGES:
- fix bug in internal table OJB_HL_SEQ, column type of MAX_ID was INTEGER but
needs BIGINT to support Long on java side

- no longer throw an exception when calling abortTransaction more than one time
or an internal rollback (by ConnectionManager on the used connection) was already done

- Add object-cache tag in standard jdbc-connection-descriptor in repository_database.xml
to enable new 'autoSync' property of ObjectCacheDefaultImpl for all operations made by
this DB connection

- Allow infinite lifetime of cached objects in ObjectCacheDefaultImpl (handle with care ;-))

- odmg-api implementation, disable restore of transient objects on transaction abort,
because we can't really restore the whole object with all references

- now using ANTLR 2.7.2 and Unicode support in OQL queries
- now using ANT 1.6.0

- RelationshipPrefetcher no longer modifies auto-retrieve setting of Relationship-Descriptor.
This may result in additional sql-queries but is safer in a multi-threaded environment.

- Added column methods to Criteria. addColumnEqualTo, addColumnIn etc. The column parameter
of these methods will not be translated during sql-generation. These methods are mainly used
for internal purpose.

BUG FIXES:

Please refer to our Bug tracking site (http://issues.apache.org/scarab/servlet/scarab/)
under http://issues.apache.org/scarab/servlet/scarab/issues/id/OJBxxx
to see details for a bug with id OJBxxx.

- fix bug in MetadataManager, when 'per thread changes' is enabled and many different
DescriptorRepository instances were used, gc can't collect unused instances because
StatementManager doesn't release references to used DescriptorRepository instances.

- fix bug in ObjectCacheDefaultImpl, it doesn't make sense to use a soft reference
for the object wrapper of the cached object. Use a softreference for the cached
object instead

- fix bug in CacheDistributor (internal used ObjectCache implementation), each CacheDistributor
instance have to held ObjectCache instances for the associated PersistenceBroker instance

- fix documentation bug in ObjectCacheDefaultImpl, property 'timeout' use
seconds instead of milliseconds

- fix bug in new DList implementation (DListImpl_2), materialization of objects
failed when DList instance itself was persisted (using db.bind(...)) and
re-loaded within tx, but iteration over the list values was done outside of the tx

- The bogus ODMG Distributed Lockmanagement feature has been replaced by a new
  Servlet based LockServer. Transaction isolation should now work properly even
  accross a cluster of JVMs.

- fix bug in MtoNCollectionPrefetcher, multi-key handling is now supported.

- ClassCastException with ManageableCollection implementations on m:n relation.
m:n relation now can handle ManageableCollection collection classes in a correct way,
the collection class no longer needs to implement java.util.Collection.

KNOWN ISSUES:
- odmg-api: It is not possible to exchange objects in 1:n references.
E.g. two objects with 1:n reference, each with one reference object, obj_1{ref_1}
and obj_2{ref_2}. Lock objects and exchange the references in collection obj_1{ref_2}
and obj_2{ref_1} and commit.


---------------------------------------------------------------------
Release 1.0 rc5
---------------------------------------------------------------------

This is the last planned rc before the 1.0 release. If there are no major
bugs this release will be relabled as 1.0 after two weeks.

NEW FEATURES:
- With this release we are feature complete for the 1.0 release!
For 1.0 you should not expect more features to be added.

NOTES:
- slight changes in repository.dtd, OJB.properties were made

- internal kernel interface method signature changed:
in JdbcAccess two method signatures change
in StatementManagerIF one method signature change
These changes are necessary to fix a "design bug" in handling
of jdbc type metadata. See discussion on dev-list "[VOTE] Design bug fixed - check in?"

- ObjectCache implementation classes constructor arguments change. We
now pass a Properties argument too. Allows to set configuration properties
for each ObjectCache instance

- changed the JDORI plugin to now use the latest 1.0.1 version of the JDO reference implementation.

- OJB is now very strict in handling RsIterator instances. RsIterator is
bound very closely to the used PersistenceBroker instance.
Thus if you do a
PersistenceBroker#close
PersistenceBroker#commitTransaction
PersistenceBroker#abortTransaction
call, the current <tt>RsIterator</tt> instance resources will be cleaned up automatic
and invalidate current instance.

CHANGES:
- add possibility to declare ObjectCache implementation on class-descriptor
and jdbc-connection-descriptor level (means per class and per database connection) too

- add a new interface called org.apache.ojb.odmg.TransactionExt
to make additional proprietary methods available for user by
casting org.odmg.Transaction to TransactionExt

- behaviour of org.odmg.Transaction#checkpoint() changed. Now the
database transaction was commited when checkpoint was called, seems this
is more in unison with ODMG spec:
" Calling checkpoint commits persistent object modifications made within the
transaction since the last checkpoint to the database."
If you want to flush persistent object
modifications made within the ODMG transaction to the underlying database
transaction without commit the changes (old behaviour of checkpoint()),
please cast Transaction to TransactionExt and use new method flush().

- make odmg collections pluggable via OJB.properties file

- Restructuring and further enhancements of the documentation.


BUG FIXES:

Please refer to our Bug tracking site under
http://scarab.werken.com/scarab/issues/id/OJBxxx to see details for a bug
with id OJBxxx.

- fix ClassLoader problem when merging DescriptorRepository instances

- fixed the JDORI problems with loading object via extend based queries.
  now objects are brought under JDO control and equipped with a statemanager
  in the load process.

---------------------------------------------------------------------
Release 1.0 rc4
---------------------------------------------------------------------

NEW FEATURES:
- With this release we are feature complete for the 1.0 release!
For 1.0 you should not expect more features to be added.
- added a XDoclet OJB Module that allows to generate repository.xml
  and DDL from tagged Java sources to the "contrib" distribution.
- added an OJB Taglib that allows to use OJB directly from JSP pages
  to the "contrib" distribution.

NOTES:
- update used xdoclet version in 'ejb-examples'
(see section deployment/'Build the OJB sample session beans')
to version 1.2xx

CHANGES:
- Enhance MetadataManager, add new methods, simplify handling
of different persistent object metadata profiles (means different
org.apache.ojb.broker.metadata.DescriptorRepository instances) at
runtime
- Provide a PUBLIC doctype declaration for the repository.xml
- add a convenience Constructor to QueryByCriteria that
  builds a query selecting all instances of a class
- add isClosed() method to PersistenceBroker interface to
check if a broker instance was already closed.
- make OJB more strict in handling closed PersistenceBroker
instances. Now it's not any longer possible to use closed
PersistenceBroker instances.
- Package ...metadata.fieldaccess refactored/changed. Now
all PersistentField implementations support 'nested fields'.
See new entries in OJB.properties file.
- added new sections to documentation


BUG FIXES:
- fix several bugs in documentation
- fixed writing to memo columns in MS Access
- fixed optimistic locking problem
- fixed jdo tutorial setup

Please refer to our Bug tracking site under
http://scarab.werken.com/scarab/issues/id/OJBxxx to see details for a bug
with id OJBxxx.


---------------------------------------------------------------------
Release 1.0 rc3
---------------------------------------------------------------------

NEW FEATURES:
- With this release we are feature complete for the 1.0 release!
For 1.0 you should not expect more features to be added.


NOTES:
- Intern table OJB_HL_SEQ (used by SequenceManagerHighLowImpl)
  has changed! Now need an additional column 'VERSION' of type INTEGER
  When updating from earlier version drop/reinstall table with additional field
- PersistenceBrokerAware interface changed, methods beforeStore and afterStore
  split into xxxInsert/xxxUpdate methods, all methods now expect an PersistenceBroker
  argument
- PersistenceBrokerListener interface replaced by PBStateListener
- ManageableCollection interface add new method afterStore(...)
- Remove obsolete method 'invalidate(Identity oid)' from PB-api


CHANGES:
- enhanced PB-api listener classes
- unified handling of events in PB-api
- adapt sequence manager implementations using 'long' instead 'int' keys
- SequenceManagerHighLowImpl now use optimistic locking to support sequencing
  across different JVM (except j2ee conform app server, see sequence manager doc).
- Add new attribute 'autoNaming' to sequence manager implementations. Now
  the implementations allow to switch on/off auto-generating of sequence names
- merging the three package o.a.ojb.broker.ta, .singlevm and .server into
  a new package o.a.ojb.broker.core

BUG FIXES:
- fixed problems with the JDO FieldManager to work with 1:1, 1:n and m:n
  associations
- OJB143, OJB152, OJB153, OJB156, OJB158, OJB159, OJB161

Please refer to our Bug tracking site under
http://scarab.werken.com/scarab/issues/id/OJBxxx to see details for a bug
with id OJBxxx.


---------------------------------------------------------------------
Release 1.0 rc2
---------------------------------------------------------------------

NEW FEATURES:
- With this release we are feature complete for the 1.0 release!
For 1.0 you should not expect more features to be added.


CHANGES:
- improve cache package. Now the used cache is no longer a singleton.
- Add CacheFilter interface to allow easy pre-filtering of cache methods.
- Add Support for anonymous fields
- Several improvements to SQL generation
- improved handling of optimistic lock exceptions in ODMG


BUG FIXES:
- lots

---------------------------------------------------------------------
Release 1.0 rc1
---------------------------------------------------------------------

NEW FEATURES:
- With this release we are feature complete for the 1.0 release!
For 1.0 you should not expect more features to be added.


CHANGES:
- improved handling for non-distributable jars
- MetadataManager: new methods, allow merge of new ConnectionDescriptor and RepositoryDescriptor
(read from any repository file at runtime) with the existing ones.
- add custom attribute support for DescriptorRepository and JdbcConnectionDescriptor.


BUG FIXES:

- repaired tutorial applications
- fixed several bugs in the JDO RI plugin
- compilation works for jdk 1.2 and 1.3 again
- connection problem when using PBKey only with jcdAlias name (e.g. PBKey("myAlias")),
now the user and password set in the repository file was used.
- add jcdAlias support for tutorial 1+2
