NAME

DBIx::Class::DeploymentHandler::DeployMethod::SQL::Translator - Manage your SQL and Perl migrations in nicely laid out directories

DESCRIPTION

This class is the meat of DBIx::Class::DeploymentHandler. It takes care of generating serialized schemata as well as sql files to move from one version of a schema to the rest. One of the hallmark features of this class is that it allows for multiple sql files for deploy and upgrade, allowing developers to fine tune deployment. In addition it also allows for perl files to be run at any stage of the process.

For basic usage see DBIx::Class::DeploymentHandler::HandlesDeploy. What's documented here is extra fun stuff or private methods.

DIRECTORY LAYOUT

Arguably this is the best feature of DBIx::Class::DeploymentHandler. It's spiritually based upon DBIx::Migration::Directories, but has a lot of extensions and modifications, so even if you are familiar with it, please read this. I feel like the best way to describe the layout is with the following example:

 $sql_migration_dir
 |- _source
 |  |- deploy
 |     |- 1
 |     |  `- 001-auto.yml
 |     |- 2
 |     |  `- 001-auto.yml
 |     `- 3
 |        `- 001-auto.yml
 |- SQLite
 |  |- downgrade
 |  |  `- 2-1
 |  |     `- 001-auto.sql
 |  |- deploy
 |  |  `- 1
 |  |     `- 001-auto.sql
 |  `- upgrade
 |     |- 1-2
 |     |  `- 001-auto.sql
 |     `- 2-3
 |        `- 001-auto.sql
 |- _common
 |  |- downgrade
 |  |  `- 2-1
 |  |     `- 002-remove-customers.pl
 |  `- upgrade
 |     `- 1-2
 |     |  `- 002-generate-customers.pl
 |     `- _any
 |        `- 999-bump-action.pl
 `- MySQL
    |- downgrade
    |  `- 2-1
    |     `- 001-auto.sql
    |- initialize
    |  `- 1
    |     |- 001-create_database.pl
    |     `- 002-create_users_and_permissions.pl
    |- deploy
    |  `- 1
    |     `- 001-auto.sql
    `- upgrade
       `- 1-2
          `- 001-auto.sql

So basically, the code

 $dm->deploy(1)

on an "SQLite" database that would simply run "$sql_migration_dir/SQLite/deploy/1/001-auto.sql". Next,

 $dm->upgrade_single_step([1,2])

would run "$sql_migration_dir/SQLite/upgrade/1-2/001-auto.sql" followed by "$sql_migration_dir/_common/upgrade/1-2/002-generate-customers.pl", and finally punctuated by "$sql_migration_dir/_common/upgrade/_any/999-bump-action.pl".

".pl" files don't have to be in the "_common" directory, but most of the time they should be, because perl scripts are generally database independent.

Note that unlike most steps in the process, "initialize" will not run SQL, as there may not even be an database at initialize time. It will run perl scripts just like the other steps in the process, but nothing is passed to them. Until people have used this more it will remain freeform, but a recommended use of initialize is to have it prompt for username and password, and then call the appropriate "CREATE DATABASE" commands etc.

Directory Specification

The following subdirectories are recognized by this DeployMethod:

"_source": This directory can contain the following directories:

"deploy": This directory merely contains directories named after schema versions, which in turn contain "yaml" files that are serialized versions of the schema at that version. These files are not for editing by hand.

"_preprocess_schema": This directory can contain the following directories:

"downgrade": This directory merely contains directories named after migrations, which are of the form "$from_version-$to_version". Inside of these directories you may put Perl scripts which are to return a subref that takes the arguments "$from_schema, $to_schema", which are SQL::Translator::Schema objects.
"upgrade": This directory merely contains directories named after migrations, which are of the form "$from_version-$to_version". Inside of these directories you may put Perl scripts which are to return a subref that takes the arguments "$from_schema, $to_schema", which are SQL::Translator::Schema objects.

A typical usage of "_preprocess_schema" is to define indices or other non-DBIC type metadata. Here is an example of how one might do that:

The following coderef could be placed in a file called _preprocess_schema/1-2/001-add-user-index.pl

 sub {
    my ($from, $to) = @_;

    $to->get_table('Users')->add_index(
       name => 'idx_Users_name',
       fields => ['name'],
    )
 }

This would ensure that in version 2 of the schema the generated migrations include an index on "Users.name". Frustratingly, due to the nature of SQL::Translator, you'll need to add this to each migration or it will detect that it was left out and kindly remove the index for you.

An alternative to the above, which is likely to be a lot less annoying, is to define such data in your schema directly, and only change it as you need to:

 package MyApp::Schema::Result::User;

 #[...]

 sub sqlt_deploy_hook ( $self, $sqlt_table ) {
    $sqlt_table->add_index(name => 'idx_Users_name', fields => [ 'name' ]);
 }

$storage_type: This is a set of scripts that gets run depending on what your storage type is. If you are not sure what your storage type is, take a look at the producers listed for SQL::Translator. Also note, "_common" is a special case. "_common" will get merged into whatever other files you already have. This directory can contain the following directories itself:

"initialize": If you are using the "initialize" functionality, you should call initialize() before calling "install". This has the same structure as the "deploy" subdirectory as well; that is, it has a directory for each schema version. Unlike "deploy", "upgrade", and "downgrade" though, it can only run ".pl" files, and the coderef in the perl files get no arguments passed to them.
"deploy": Gets run when the schema is "deploy"ed. Structure is a directory per schema version, and then files are merged with "_common" and run in filename order. ".sql" files are merely run, as expected. ".pl" files are run according to "PERL SCRIPTS".
"upgrade": Gets run when the schema is "upgrade"d. Structure is a directory per upgrade step, (for example, "1-2" for upgrading from version 1 to version 2,) and then files are merged with "_common" and run in filename order. ".sql" files are merely run, as expected. ".pl" files are run according to "PERL SCRIPTS".
"downgrade": Gets run when the schema is "downgrade"d. Structure is a directory per downgrade step, (for example, "2-1" for downgrading from version 2 to version 1,) and then files are merged with "_common" and run in filename order. ".sql" files are merely run, as expected. ".pl" files are run according to "PERL SCRIPTS".

Note that there can be an "_any" in the place of any of the versions (like "1-2" or 1), which means those scripts will be run every time. So if you have an "_any" in "_common/upgrade", that script will get run for every upgrade.

PERL SCRIPTS

A perl script for this tool is very simple. It merely needs to contain an anonymous sub that takes a DBIx::Class::Schema and the version set as it's arguments.

A very basic perl script might look like:

 #!perl

 use strict;
 use warnings;

 use DBIx::Class::DeploymentHandler::DeployMethod::SQL::Translator::ScriptHelpers
    'schema_from_schema_loader';

 schema_from_schema_loader({ naming => 'v4' }, sub {
   my $schema = shift;

   # [1] for deploy, [1,2] for upgrade or downgrade, probably used with _any
   my $versions = shift;

   $schema->resultset('Users')->create({
     name => 'root',
     password => 'root',
   })
 })

Note that the above uses "schema_from_schema_loader" in DBIx::Class::DeploymentHandler::DeployMethod::SQL::Translator::ScriptHelpers. Using a raw coderef is strongly discouraged as it is likely to break as you modify your schema.

ATTRIBUTES

ignore_ddl

This attribute will, when set to true (default is false), cause the DM to use SQL::Translator to use the "_source"'s serialized SQL::Translator::Schema instead of any pregenerated SQL. If you have a development server this is probably the best plan of action as you will not be putting as many generated files in your version control. Goes well with with "databases" of "[]".

force_overwrite

When this attribute is true generated files will be overwritten when the methods which create such files are run again. The default is false, in which case the program will die with a message saying which file needs to be deleted.

schema

The DBIx::Class::Schema (required) that is used to talk to the database and generate the DDL.

storage

The DBIx::Class::Storage that is actually used to talk to the database and generate the DDL. This is automatically created with "_build_storage".

sql_translator_args

The arguments that get passed to SQL::Translator when it's used.

script_directory

The directory (default 'sql') that scripts are stored in

databases

The types of databases (default "[qw( MySQL SQLite PostgreSQL )]") to generate files for

txn_prep

This attribute will, when set to false (default is true), cause the DM to generate SQL without enclosing "BEGIN" and "COMMIT" statements.

The (current) default behavior is to create DDLs wrapped in transactions and to remove anything that looks like a transaction from the generated DDLs later when running the deployment.

Since this default behavior is error prone it is strictly recommended to set the "txn_prep" attribute to false and remove all transaction statements from previously generated DDLs.

AUTHOR

Arthur Axel "fREW" Schmidt <frioux+cpan@gmail.com>

COPYRIGHT AND LICENSE

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.