DataNucleus AccessPlatform 2.1 Release Notes

Fix to bug in determination of inheritance level of application identity object

Fix couple of places where code not thread-safe

Fix to ConnectionFactory2 usage

JPA : Support for fetch groups (DataNucleus extension)

JPA : few minor corrections to JPA2 handling

Drop "attachPolicy" and provide complete attachment in default scenario

Couple of fixes to result handling in in-memory query evaluator

RDBMS : Allow column reuse for multiple fields of a class

RDBMS : Fix to use of subclass-table with 1-1 bidirectional relations

RDBMS : Fix to use of lengths with BLOB/CLOB on MySQL

RDBMS : Fix to index/FK creation on Derby to avoid warning messages

JPA : Outline implementation of JPA2 lock methods

JPA : Support use of @PrimaryKeyJoinColumn for 1-1s

Improved merge of metadata information when sequence or cascade info specified in annotations

Fix to PMF startup process so that persistence properties are registered before enablement of some services

RDBMS : fix to use of positional parameters

RDBMS : fix to validation of candidate keys, that prevented creation of some unique keys under some circumstances

RDBMS : fix to List.get() when used in result clause of JDOQL query

RDBMS : fix to creation of sequences with SchemaTool to take extension props

JPA : fix to case where sequence name not set

JPA : fix to EM.getReference to not validate the object

JPA : fix to use of @Enumerated with @Column

Enhancer : fix to runtime enhancement, class loading problem

Enhancer : fix to error message of override of invalid field

Cache : add option to allow retention of cached objects after close of PMF/EMF, for the case where used by other instances

Changed method for instantiating objects using application identity when returned by queries, reducing the memory and CPU requirement

RDBMS : Ability to specify global default for sequence/increment value generator cache size

RDBMS : Fix to BLOB/CLOB handling with Derby 10.6+

RDBMS : Full support for HSQLDB 2.0

RDBMS : Fix to query String.startsWith/endsWith on H2

RDBMS : Fix to query String.substring(int, int) on Derby

RDBMS : Fix to BLOB/CLOB handling in queries with MSSQL

RDBMS : Support for batched INSERTs, so you re-run all of your benchmarks ;-)

RDBMS : SchemaTool now creates necessary auto-start table, sequence table, and sequences

JSON : Fix to HTTP Error 409 handling

Changed to use JDO3 jar

Support for persistence to GoogleStorage

Support for persistence of some Google Collections classes

Added initial version of a generic query optimiser, removing redundant variable clauses

Fix to JPA inheritance strategy "JOINED" when having subclasses with discriminator

Fix to persist of new object with collection that contains detached objects

Fix to persist of new object with related detached compound identity chain

Optimise out some SCO collection operations such as add+delete of same object in succession

Support persist of enums as arbitrary value

JPA2 : Support for orphan removal

JPA2 : Support for JPQL "INDEX" keyword

RDBMS : Fix to setting of BLOB/CLOB on bidirectional relations for Oracle

RDBMS : Fix to SQL table namer to allow for unioned statements when choosing table name

RDBMS : Support for use of BoneCP connection pools

Change some Maven dependencies to allow use with JDO3

Fix to generic compilation for "param.field" and use of subquery candidate

Extend in-memory query evaluation result clause handler to allow variable expression and more

Fix to in-memory query evaluation for String.matches when the string is null

RDBMS : Fix to cascade delete on 1-1 when there is a FK present (so leave to datastore)

RDBMS : Fix to Derby syntax for CROSS JOIN pre-10.6

RDBMS : Fix to Derby usage with JDOQL2 to restrict use of CLOB/BLOC with DISTINCT/UNION

RDBMS : Add ability to query in datastore when candidate collection being input

RDBMS : Fix to Map.containsKey/containsValue to allow for contained variable already having a table in the statement

RDBMS : Fix to use of subqueries with parameters linking back to the main query

RDBMS : Fix to use of input positional parameters when one is used multiple times in the statement

RDBMS : Fetching optimisation when fetching a field and taking the opportunity to load any other non-relation or 1-1/N-1 unloaded fields

RDBMS : Fix to SQL table namers to allow for multiple levels of subquery

RDBMS : Fix to Collection.contains(Enum)

RDBMS : Fix to use of parameter with null value in comparisons

RDBMS : Fix to use of explicit parameters in subquery

RDBMS : Fix to generation of "count" query to get size so it takes into account DISTINCT in the main query

Excel : Support for persistence of primitive wrapper types

Excel : Support for persistence of embedded PC fields (Guido)

Spatial : Added support for all Oracle/PostGIS/MySQL specific methods

Spatial : Fix to handling of user object in Jts objects when querying

Spatial : Addition of remaining methods for querying

Add ability to skip the check on unused input parameters to queries

Fix issue with detachment when finding an object already detached but not far enough

Add support for compilation of queries with parameters in subqueries linking back to the main query

Fix to L2 caching of StringBuffer fields to cache a copy

Fix to handle input query with special characters (tab, eol, etc)

Extend query compilation to store whether the query result is distinct

Support specification of interface fields where there isn’t currently an available implementation

Fix to use of dynamic fetch groups in multiple threads

Fix to support non-standard SQL statements (stored procs, statements not starting with SELECT/INSERT/UPDATE/MERGE/DELETE)

Fix to the query in-memory evaluator to support aggregate names in upper or lower case

Fix to a couple of situations where a versioned object can be returned to the user without the version set

Fix to reset the version of an object on rollback()

Fix to SCOListIterator

RDBMS : Fix to generation of indexes for reference fields when having a FK per implementation so that the index is created for the implementations FK only

RDBMS : Fix JDOQL2 to not select CLOB fields when using UNION or DISTINCT for Derby

RDBMS : Fix JDOQL2 to add parentheses around ORed clauses

RDBMS : Fix JDOQL2 to pick the correct statement when using methods in subqueries

RDBMS : Fix JDOQL2 "alpha-namer" to handle more than 26 table groups

RDBMS : Fix JDOQL2 to handle comparisons of interface field with an implementation

RDBMS : Support use of CHECK constraint when persisting boolean as numeric

RDBMS : Change HSQLDB use of LIMIT to append to the end of the statement

RDBMS : Support for use of SQLXML type with DB2 (contrib)

RDBMS : Fix to JDOQL2 for use of input parameters with multiple columns

JPA : Add support for JPQL "case" expressions

JPA : Add support for generation of JPA2 static metamodel classes

JPA : Change to handling of "columnDefinition" to expect the type to be included

Enhancer : when using Java6 and having the enhancer in the CLASSPATH default to not enabling the annotation processor to enhance the class(es) being compiled

Addition of plugin for persistence of javax.time (JSR0310) types

Fix to a typo in the use of ConnectionFactory when defining the datastore as a DataSource

Cleanup to process of creation of L1/L2/Query caches

Change generic query compilation expression tree bidirectional for ease of navigation

Support generic query compilation of {param}.field

Enforce JDO2 spec rules for JDOQL result/grouping/ordering clauses validity

Fix to generic query compilation of {variable}.field

Fix to generic query compilation of cast expressions

Added checks on invalid use of keywords during generic query compialtion

Fix to generic query compilation of {array}.length

Fix to use of Extent to close all iterators correctly

FIx to in-memory evaluation of queries to better handle date equality

Fix to getParent() of class/package metadata of JDO Metadata API

Fix to use of set() on ListIterator obtained from list wrapper

Fix to persistence of object with reference to the same object

RDBMS : change to use "JDOQL2" implementation by default. Use "JDOQL-Legacy" if wanting the old JDOQL implementation.

RDBMS : having expression that are not boolean should throw XXXUserException for JDOQL2

RDBMS : fix to use of parameters for range and passed in as unnamed to API

RDBMS : fix to caching of queries when using String.matches with input parameter for JDOQL2

RDBMS : fix to not create indexes for serialised field (problem on MySQL)

RDBMS : support override of datastore mapping by user plugins

RDBMS : fix to only use CROSS JOIN with Derby from v10.6 onwards

RDBMS : fix to use of Boolean expression/literals for JDOQL2

RDBMS : fix to candidate query on persistent interface candidate for JDOQL2

RDBMS : add support for persistent object "identity" as input parameter for JDOQL2

RDBMS : fix to numeric type comparisons for JDOQL2

RDBMS : fix to use of parameter in result clause when null for JDOQL2

RDBMS : add support for Cast expression in result clause for JDOQL2

RDBMS : add support for array of expressions in JDOQL2

RDBMS : add support for joins to interface implementations in JDOQL2

RDBMS : add support for subquery.contains() in JDOQL2

RDBMS : fix handling of DISTINCT in JDOQL2

RDBMS : make use of Query imports when resolving classes in instanceof/cast for JDOQL2

RDBMS : fix use of navigation of N-1 relation via join table in JDOQL2

RDBMS : Delete "JPQL-Legacy" implementation

JodaTime : fix handling of null fields

JPA : upgrade to use JPA2 "final" jar API

Much internal refactoring has also been performed to remove unneeded components or to simplify the API’s for things needed in the 2.1 timeline. Any use of internal API’s by applications will likely need changes.