<ul><li>The default jar file is now compiled for Java 6.
<ul><li>The default jar file is now compiled for Java 6.
</li><li>The new jar file will probably not end up in the central Maven repository
</li><li>The new jar file will probably not end up in the central Maven repository
in the next few weeks because Sonatype has disabled automatic synchronization
in the next few weeks because Sonatype has disabled automatic synchronization
...
@@ -307,78 +311,5 @@ Change Log
...
@@ -307,78 +311,5 @@ Change Log
setting "defrag_always" was used.
setting "defrag_always" was used.
</li></ul>
</li></ul>
<h2>Version 1.3.158 (2011-07-17)</h2>
<ul><li>Fulltext search (native): after re-opening the connection once the fulltext index was created,
a transaction rollback did not roll back the modifications in the index.
</li><li>Support for ROW_NUMBER() OVER() as an alias for ROWNUM()
for compatibility with PostgreSQL and Apache Derby.
</li><li>Issue 325: a deadlock in the fulltext search could occur if the connection that was
used while initializing the fulltext search was later used to query the database.
This was actually not only a problem for fulltext search, but also for other triggers
that used the initializing connection later on.
</li><li>DATABASE_TO_UPPER: when set to false, all identifier names (table names, column names)
are case sensitive (except aggregate, built-in functions, data types, and keywords).
This is for improved compatibility with MySQL and PostgreSQL.
</li><li>When upgrading from an older 1.3.x version to version 1.3.157, when using BLOB or CLOB data,
the database could only be opened once.
</li><li>SUM on a boolean expression will now count the number of 'true' rows.
</li><li>Simplified test cases using assertThrows(...).
</li><li>A Java function that returns a result set that was generated using the provided connection
could cause an internal error in some cases. Now such result sets are copied early on.
</li><li>Data type RESULT_SET: information_schema.function_aliases returned type 0; now type -10 is returned (the same as Oracle: OracleTypes.CURSOR = -10).
</li><li>Data type RESULT_SET: tables with columns of type RESULT_SET could be created, but no data could be persisted in this column.
</li><li>H2 Console: improved system tray icon for Mac OS X (transparent background).
</li><li>String.toUpperCase() was used a few places, which is problematic when using the Turkish locale.
The method has been replaced with toUpperCase(Locale.ENGLISH) to solve such problems.
</li><li>Shell tool: the built-in command "distinct" has been removed
(use the SQL statements "show tables" / "show columns from tableName" instead).
The result set formatting has been improved.
</li><li>MERGE: if a unique key was violated (but not the primary key or the key columns of the merge itself),
the wrong exception was thrown.
</li><li>IBM DB2 and Apache Derby compatibility: support the pseudo-table SYSIBM.SYSDUMMY1.
</li><li>ROWNUM() did not work in combination with IN(..). The following query did not work as expected:
select * from (select rownum r from test) where r in (1, 2).
</li><li>H2 Console autocomplete: the autocomplete feature didn't support quoted names.
</li><li>It is now longer allowed to create an index on a CLOB or BLOB column
(except for in-memory databases or indexes), because recovery doesn't work
on such columns. Fulltext indexes on such column are still supported of course.
</li><li>The PUBLIC schema could be renamed, which could result in a database that
can't be opened normally. Now it can no longer be renamed.
The schema INFORMATION_SCHEMA can also no longer be renamed.
</li><li>Server mode: the problem with server and client using different
timezones with different daylight saving rules has been solved
(this only applies when both the client and the server use this or a newer version).
</li><li>There is a new system property "h2.storeLocalTime" to avoid
problems opening a database file in a timezone that has different
daylight saving rules (see also below).
</li><li>Date, time, and timestamp data type processing has been re-implemented.
Time now supports nanoseconds resolution and now now supports a wider range
(negative and large values), similar to PostgreSQL.
</li><li>SQL statements with a non-breaking space were considered invalid.
</li></ul>
<h2>Version 1.3.157 (2011-06-25)</h2>
<ul><li>The syntax for CREATE TABLE ... AS SELECT ... has been changed:
options such as NOT PERSISTENT are now before the query (AS SELECT), because otherwise
they might be interpreted as a part of the query.
</li><li>With version 1.3.156, the DB2 mode could not be used with CLOB and BLOB data.
</li><li>When upgrading from an older 1.3.x version to version 1.3.156,
CLOB and BLOB data was lost in many cases.
</li><li>The optimization for COUNT(..) on columns that are not nullable
was also used for COUNT(DISTINCT ..), which is incorrect.
</li><li>PgServer: non-admin users could not open a database.
</li><li>Non-admin users could not open a database using a mode (MODE=xxx)
if the database was already open, even if the mode matched.
</li><li>The SQL state of the following exceptions has been changed:
From 90009, 90010, 90011 to 22007: "Cannot parse DATE/TIME/TIMESTAMP constant ...".
</li><li>There is a problem when opening a database file in a timezone that has different
daylight saving rules: the time part of dates where the daylight saving doesn't match
will differ. This is not a problem within regions that use the same rules (such as, within
USA, or within Europe), even if the timezone itself is different. As a workaround, export the
database to a SQL script using the old timezone, and create a new database in the new
timezone. The same problem occurs when accessing a database over TCP/IP if the client and server
timezones use different rules, if an old server or client is used.
