NAME
    DBIx::Migration - Seamless database schema up- and downgrades

SYNOPSIS
      use DBIx::Migration ();

      my $m = DBIx::Migration->new(
        dsn => 'dbi:SQLite:~/Projects/myapp/db/myapp.db'
      );

      # Get current version from database
      my $version = $m->version;

      # Set directory before applying migrations
      $m->dir( '~/Projects/myapp/db/migrations' );

      # Migrate database to version 1
      $m->migrate( 1 );

      # Migrate database to the latest version
      $m->migrate;

      # ~/Projects/myapp/db/migrations/schema_1_up.sql
      CREATE TABLE foo (
        id INTEGER PRIMARY KEY,
        bar TEXT
      );

      # ~/Projects/myapp/db/migrations/schema_1_down.sql
      DROP TABLE foo;

      # ~/Projects/myapp/db/migrations/schema_2_up.sql
      CREATE TABLE bar (
        id INTEGER PRIMARY KEY,
        baz TEXT
      );

      # ~/Projects/myapp/db/migrations/schema_2_down.sql
      DROP TABLE bar;

DESCRIPTION
    This module provides seamless database schema up- and downgrades. The
    implementation is based on migrations. A migration is a ".sql" script.
    Although not mandatory the script name begins with a prefix like for
    example "schema_". It follows a version number that is a positive
    integer. After an "_" (underscore) character the script name ends with
    the migration type that is either "up" or "down". Migrations are stored
    in a directory and are applied in order to a database. Usually the
    version number of the first migration is 1. The version numbers of the
    other migrations have to be ascending without gaps.

    During processing the content of each migration is read with the
    "binmode" of ":raw" into a scalar. The content is split into sections
    using the default SQL delimiter ";" (semicolon). Each section is
    executed independently. All related sections are encapsulated in a
    database transaction. If a migration embeds stored logic containing one
    or more semicolons (a PostgreSQL trigger function for example), the
    "migrate()" method incorrectly splits the migration into sections,
    causing an error. You can set a different delimiter to overcome this
    problem. Add the "dbix_migration_delimiter" annotation as an SQL comment
    to the migration

      -- dbix_migration_delimiter: /
      ...

    The annotation has to be specified in the first line. The delimiter has
    to be a single printable ASCII character, excluding the space character.
    In the previous example it is the "/" (slash) character.

METHODS
    $self->BUILD( $args )
        Validate object.

    $self->debug( $debug )
        Enable/disable debug messages.

    $self->dir( $dir )
        Get/set-once directory that contains migrations.

    $self->dsn
        Get data source name.

    $self->dbh
        Get database handle.

    $self->migrate( $version )
        Migrate database to version. Returns true in case of success;
        otherwise false. If called without the version argument the latest
        migration version will be used.

    $self->password
        Get database password.

    $self->username
        Get database username.

    $self->version
        Get migration version from database. Will be "undef" if no migration
        has taken place yet. The version is stored in the table
        "dbix_migration". The name of this table may change in the future so
        don't rely on it.

SEE ALSO
    *   Liquibase endDelimiter
        <https://docs.liquibase.com/change-types/enddelimiter-sql.html>

TODOS
    *   Implement proper logging. Printing debug information to stdout
        conflicts with the stdout output of "DBIx::Migration::CLI".

AUTHOR
    Sebastian Riedel, <kraihx@gmail.com>

CONTRIBUTORS
    Dan Sully, <dan+github@sully.org>

    Marcus Ramberg, <marcus@nordaaker.com>

    Steven Jenkin, <sjenkin@venda.com>

    Sven Willenbuecher, <sven.willenbuecher@gmx.de>

COPYRIGHT
    Copyright 2004-2005 Sebastian Riedel. All rights reserved.

    This program is free software, you can redistribute it and/or modify it
    under the same terms as Perl itself.