CHANGES

Version 1.1.1 - January 1st, 2013

  * Use slf4j-api as a library dependency and slf4j-log4j12 as a unit
    testing dependency.  Using slf4j-log4j12 as a library dependency
    required that dependent libraries and applications load
    slf4j-log4j12 which defeats the purpose of SLF4J.  Closes
    https://github.com/blair/scala-migrations/issues/1 .

  * Enable unit testing on PostgreSQL when the Java
    "scala-migrations.db.vendor" property is set to "postgresql".

    Patches by Alan LaMielle.

Version 1.1.0 - December 28th, 2012
  * Migration adds #databaseVendor to get the database vendor which
    can be used for pattern matching for database specific code.

  * Add an AutoIncrement column option which is supported on Derby,
    MySQL and PostgreSQL for the SMALLINT, INTEGER and BIGINT column
    data types.  It instructs the database to provide an
    auto-incrementing default value if the application does not
    provide one when inserting a new row to the table.

    Derby:
      Only supported on SMALLINT, INT and BIGINT data types using
      Derby's GENERATED BY DEFAULT AS IDENTITY.  The alternate setting
      GENERATED ALWAYS AS IDENTITY is not used as it is not consistent
      with MySQL and PostgreSQL which permits the application to
      explicitly specify the column's value.

      http://db.apache.org/derby/docs/10.9/ref/rrefsqlj37836.html

    MySQL:
      Only supported on SMALLINT, INT and BIGINT data types using
      MySQL's AUTO_INCREMENT keyword.

      http://dev.mysql.com/doc/refman/5.5/en/create-table.html
      http://dev.mysql.com/doc/refman/5.5/en/example-auto-increment.html

    PostgreSQL:
      Only supported on SMALLINT, INT and BIGINT data types by
      replacing the data type name with SMALLSERIAL, SERIAL and
      BIGSERIAL, respectively.  Support for SMALLSERIAL is only
      available in PostgreSQL 9.2 and greater.

      http://www.postgresql.org/docs/9.2/static/datatype-numeric.html#DATATYPE-SERIAL

    Oracle:
      No support is provided in this commit as it appears that
      equivalent functionality can only be provided by using triggers.

  * Add support for MySQL:
    - MySQL database adapter used if the JDBC driver class is named
      com.mysql.jdbc.Driver.
    - For grant and revoke operations, generalize the concept of a
      user to support MySQL style accounts, e.g. `username`@`hostname`.

      An abstract User base class with two concrete subclasses,
      PlainUser and MysqlUser, was added.  String usernames,
      e.g. "foobar", used on MySQL are converted to the same user from
      localhost, e.g. `foobar`@`localhost`, as this is the most secure
      default.  If database specific usernames are required, pattern
      match on the database vendor:

        def up() {
          val user = databaseVendor match {
                       case Mysql => MysqlUser("foobar", "%")
                       case _ => PlainUser("foobar")
                     }
          ...
          ...
        }
    - Generalize the SQL that drops a drop foreign key constraint.
    - Generalize the character used to quote an identifier; MySQL uses '`'.
    - Generalize for databases which do not have table and column
      constraints, e.g. MySQL.
    - Add support for character sets having a collation.
    - The Unicode case object specifies the "utf8" character set with
      the "utf8_unicode_ci" collation.  For an explanation why this
      slower collation is used instead the "utf8_general_ci"
      collation, which is MySQL's default collation for the "utf8"
      character set, see http://stackoverflow.com/questions/766809/ .
    - Unit tests run on MySQL.

  * On Derby:
    - A limit on the BLOB data type is now reflected in the generated
      SQL passed to Derby.
    - A limit on a TIMESTAMP is now not passed to Derby since Derby
      does not support it.
    - Unit tests now use an in memory database for faster turnaround.

  * On PostgreSQL:
    - Default values on the BLOB and VARBINARY data types are now
      reflected in the generated SQL passed to PostgreSQL.

  * To support PostgreSQL's ability to grant and revoke privileges on
    schemas and its USAGE privilege, the following was done:

    - Add a new sealed base Privilege trait that is the super-trait to
      all privileges.
    - Add a new sealed SchemaPrivilege trait that extends Privilege.
    - Add a new UsagePrivilege case object that extends SchemaPrivilege.
    - The existing GrantPrivilegeType trait now also extends Privilege.
    - The existing AllPrivileges case object now also extends SchemaPrivilege.
    - Migration adds #grantSchemaPrivilege() and #revokeSchemaPrivilege()
      to add or remove privileges from a schema.

    Patches by Alan LaMielle.

  * The With.*() methods properly handle the closure argument and/or
    close() throwing exceptions; functionality inspired by Java 7's
    try-with-resources where an exception thrown by closer is not
    suppressed if the body did not throw an exception.  This ensures
    that the caller always is thrown an exception if one occurred.

    http://docs.oracle.com/javase/specs/jls/se7/html/jls-14.html#jls-14.20.3

  * Scala 2.10.0:
    - Add support for 2.10.0.
    - Explicitly type exception classes in exception catching pattern
      matches.  When catching Throwable always log it or rethrow it.
    - Fix minor bugs caught by Scala 2.10.0 compiler.

  * Code style cleanups to better match the Scala style guide:
    http://docs.scala-lang.org/style/scaladoc.html

Version 1.0.3 - September 21st, 2012
  * Make log4jdbc optional by checking for it at runtime and using it
    if it is available.  This is to workaround log4jdbc not present in
    any public Maven repositories, to make it easier for developers
    using build tools with automatic dependency resolution so they do
    not have to download log4jdbc themselves:
      http://code.google.com/p/log4jdbc/wiki/FAQ
      http://code.google.com/p/log4jdbc/issues/detail?id=19

  * For Scala 2.8.0 and greater, switch from ant/ivy to sbt to build
    the project.  In addition to reducing the number of files checked
    into source control to build the project, it also supports easy
    publishing to Sonatype's open-source repository.

Version 1.0.2 - November 8th, 2011
  * Issue #22: use java.lang.ClassLoader#getResources() instead of
    #getResource() in case there are multiple resources with the same
    package name.  Now all resources providing the package name are
    searched.

  * Issue #23: fix a bug by URL decoding resource URLs so that
    resources in directories containing one or more space characters
    can be found.

Version 1.0.1 - February 14th, 2010
  * New method Migration#alterColumn() which allows column definitions
    to be altered.

  * Initial support for building with Maven.  It only builds jars for
    JDBC 4/JDK 1.6, not for JDBC 3/JDK 1.5.  Build with ant to build
    JDBC 3/JDK 1.5 jars.

  * Usernames for Migration#grant() and #revoke() are now converted
    into the database's canonical case, e.g. lowercase or uppercase,
    and then quoted.

  * Refactor the unit tests to allow for testing with databases other
    than Derby.

  * The table names in the unit tests have been renamed to all start
    with "scala_migrations_".  This makes it easier to unit test the
    project in an existing schema.  The only table that doesn't start
    with "scala_migrations_" is "schema_migrations".

  * Lots of internal refactoring and code cleanup.

Version 1.0.0 - February 11th, 2010
  * Provide Scala 2.8.0.Beta1 support.  Code modifications that work
    under both Scala 2.7.x and 2.8.x were applied to the default
    (2.7.x) branch and merged to the scala-2.8 branch to minimize the
    differences between the branches.

  * Fix a bug in dropping indices in Oracle.

  * Index names are now quoted and potentially case converted in the
    same manner as column names.

  * Provide a addingForeignKeyConstraintCreatesIndex method in
    Migration and DatabaseAdaper that returns true if the database
    implicitly adds an index on the column that has a foreign key
    constraint added to it.  This can be used to dynamically add an
    index for those databases that don't implicitly add an index on
    foreign key columns.

  * Switch to using scaladoc instead of vscaladoc when compiling with
    Scala 2.7.x to be consistent with compilation with Scala 2.8.x.

  * Update the test suite to run against Derby 10.5.3 instead of
    10.5.1.1.

  * Switch code style to be more compliant with standard Scala code,
    specifically, remove the space before a : in type annotations.

Version 0.9.2 - November 2nd, 2009
  * Build with Scala 2.7.7.

Version 0.9.1 - September 2nd, 2009
  * Add the ability to add and drop columns in existing tables.  The
    new methods are addColumn() and removeColumn() in the Migration
    class.

Version 0.9.0 - August 1st, 2009
  * First public release.