NAME

Rose::DBx::RegistryConfig - Rose::DB with auto-registration of data sources from YAML configuration file

DESCRIPTION

Rose::DBx::RegistryConfig helps you work with data source definitions in YAML-based configuration files, supporting multiple "namespace representations." It allows you to register Rose data sources without hard-coding anything directly in source code.

MOTIVATION

Using configuration files to store data source definitions instead of putting this information (which amounts to configuration details) directly in source code (as is typically done when using Rose::DB) is a valuable convenience in general. It becomes especially valuable as the number of data sources increases.

The end goal is to cleanly organize configuration data. This is not just a matter of aesthetics. Small, self-contained configuration files reduce error and save time. They are naturally easy to maintain.

SYNOPSIS & IMPORT ARGUMENTS

#------- First create a local subclass (recommended):
package My::DB;
use base qw( Rose::DBx::RegistryConfig );
...
1;
__END__

#------- Then use it in your client code:

# Use with a DOMAIN_CONFIG file to auto-register all domains:
use My::DB
    default_domain  => 'devel',
    default_type    => 'mydb',
    domain_config   => '/path/to/DOMAIN_CONFIG';

# ...or register only a subset of the domains in DOMAIN_CONFIG:
use My::DB
    domain_config   => '/path/to/DOMAIN_CONFIG',
    target_domains  => [ qw( domain1 domain2 ) ];

# ...or just use an existing registry:
use My::DB
    registry        => $registry;   # ($registry defined at compile-time)

# ...a custom namespace representation can also be supported instead of the default:
use My::DB
    domain_config   => '/path/to/DOMAIN_CONFIG',
    parse_domain_hash_callback  => \&my_domain_parser;

# (after 'use()'ing, proceed as you would with Rose::DB...)

Rose::DBx::RegistryConfig is a specialization of Rose::DB. Understanding the basic usage of Rose::DB is essential.

Rose::DBx::RegistryConfig provides some alternative ways for working with the Rose::DB Registry. Beyond that sphere of responsibility, it behaves like Rose::DB. As with Rose::DB, Rose::DBx::RegistryConfig is intended to be subclassed.

Most interaction with the interface usually takes place via import arguments (arguments to use). However, all import arguments are optional.

Import arguments for basic class-wide settings...

default_domain

Define the class-wide default domain.

default_type

Define the class-wide default type.

Arguments for initializing the data source registry from the DOMAIN_CONFIG file are also accepted. See the arguments by the following names in conf2registry:

domain_config
target_domains
parse_domain_hash_callback

...or, mutually-exclusive to arguments dealing with DOMAIN_CONFIG:

registry

A pre-made data source registry object. This allows you to explicitly cause an existing registry to be used (NOTE that setting this argument with use constitutes the use of variable data at compile time, so the registry must be available then, e.g. by creating it in a BEGIN block).

Tip: dynamically setting import arguments

When you need to dynamically set arguments to use(), make sure that they are defined at compile time:

# Importing with dynamic arguments...
my $domain_config;
BEGIN {
    $domain_config = get_rose_dbx_domains_from_somewhere();
}
use Rose::DBx::RegistryConfig
    default_domain  => $default_domain,
    default_type    => $default_type,
    domain_config   => $domain_config;

DOMAIN_CONFIG

DOMAIN_CONFIG is a YAML file containing data source definitions. Rose::DBx::RegistryConfig interprets the following namespace representation by default:

# an example domain specifically for a collection of similar databases:
dev-private:
    defaults:
        driver:     mysql
        host:       dbhost
        username:   me
        password:   foo
    DATASET_A:
    DATASET_B:
    DATASET_C:
    DATASET_D:
    DATASET_E:
---
# another domain:
otherdomain:
    defaults:
        somemethod:     somevalue
    sometype:
        othermethod:    othervalue
    

This namespace representation is used as the Rose::DBx::RegistryConfig default because the Rose::DB default representation leads to a large amount of redundant information for configurations that involve many similar databases.

Note especially the following about this namespace representation:

  • The standard namespace representation used by Rose for ROSEDBRC and auto_load_fixups is the same as this one except it does not have a 'defaults' pseudo-type and is more explicit. This means that either representation can be used with Rose::DBx::RegistryConfig.

  • DOMAIN_CONFIG must consist of a sequence of Rose domains, which each contain Rose types and their definitions.

  • In addition to the normal types of Rose::DB, the 'defaults' pseudo-type is recognized. The domain parser assumes the default value for each attribute that is not defined for a type. Thus, in the above example, all DATASET data sources will have the value 'mysql' for the driver attribute, 'dbhost' for the host, etc.

  • The DATASET_X type names have no 'database' method/value even though the defaults do not provide a database attribute. Where does the database name come from? The default domain parser knows to use the type name (DATASET_X) as the database name if the attribute is omitted.

  • This representation is also supported in the "fix-ups" file used for the ROSEDBRC/auto_load_fixups feature.

Alternative representations may be handled using a domain parser, but NOTE the following restriction: DOMAIN_CONFIG should consist of a set of domain names (the top-level keys). The values can define types in any way desired, as long as it's YAML. If this is too restrictive then set the registry explicitly.

CLASS METHODS

conf2registry

my $reg = Rose::DBx::RegistryConfig->conf2registry(
    domain_config   => $domain_config,
    target_domains  => [ 'd1', 'd2', 'd3' ],
    parse_domain_hash_callback  => \&my_domain_parser,
);

Parse DOMAIN_CONFIG and use its contents to create a data source registry. This allows data source definitions to be kept in a file instead of in source code, which is encouraged because data source definitions are, conceptually, configuation data.

domain_config

(Required)

This allows you to specify the path to DOMAIN_CONFIG. With this import argument, a data source registry is automatically created for this class based on the data sources defined in your DOMAIN_CONFIG file.

target_domains

An array of domain names for auto_registration. This defines the set of domains which will be auto-registered. All other domains will be excluded. This lets you ensure that only a subset of the data source definitions in DOMAIN_CONFIG will be registered, which might be useful if DOMAIN_CONFIG is being used for multiple tasks or multiple apps.

parse_domain_hash_callback

A subroutine reference to a caller-defined alternative to the default domain parser. It is called with the same arguments as the default domain parser and is responsible for the same task. It differs only in that it is used to implement an alternative namespace representation.

parse_domain

This method is the class-wide default domain parser, responsible for creating a set of registry entries from a data structure that represents a domain. It recognizes the class-wide default data source namespace representation.

The domain parser is called automatically for each domain in DOMAIN_CONFIG. It must interpret a given domain data structure, which should represent a single domain in the data source registry, as a set of registry entries. These entries are added to the provided registry object, which is finally returned.

registry

(Required. Must be a descendant of Rose::DB::Registry.)

A Rose::DB::Registry object to operate on.

domain_name

(Required)

The name of the domain to be registered.

domain_hashref

(Required)

Data structure containing the definition of the domain to be interpreted.

SUBCLASSING

See the notes about derived classes in the Rose::DB documentation.

Additionally, subclasses may implement a class-wide default data source namespace representation by overriding the default domain parser.

Note also that if your subclass is to support your new namespace representation for the ROSEDBRC/auto_load_fixups feature (doing this where applicable is a good idea for consistency -- it would be best to use the same representation for ROSEDBRC and DOMAIN_CONFIG), you also need to override load_yaml_fixups_from_file.

auto_load_fixups

This method overrides auto_load_fixups in Rose::DB. This is done so that alternative namespace representations can be used within ROSEDBRC. Aside from supporting alternative representations, this method functions in the same way. See load_yaml_fixups_from_file.

load_yaml_fixups_from_file

This method is called by auto_load_fixups when a file is being used to indicate "fix-up" information. Subclasses should override it if an alternative namespace representation is being used.

It is called as a class method with one (additional) argument: the name of the ROSEDBRC file containing fix-up data.

DIAGNOSTICS

DOMAIN_CONFIG file '...' not found

The supplied path for DOMAIN_CONFIG was not found.

param '...' required

Missing a required subroutine parameter.

param '...' must be a <class> object

A given subroutine parameter is not an object of the required type (<class>).

CONFIGURATION AND ENVIRONMENT

See Rose::DB. Rose::DBx::RegistryConfig adds the following features that impact configuration/environment:

DOMAIN_CONFIG

DEPENDENCIES

Carp

YAML

Rose::DB

LICENSE AND COPYRIGHT

Copyright (c) 2009 Karl Erisman (karl.erisman@icainformatics.com), ICA Informatics. All rights reserved.

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

ACKNOWLEDGEMENTS

Thanks to ICA Informatics for providing the opportunity for me to develop and to release this software. Thanks also to John Ingram for ideas about the simplified representation of the default DOMAIN_CONFIG, which is helping us reduce the complexity of our configurations significantly.

Thanks also to John Siracusa for the Rose family of modules and for providing guidance in the form of answers to questions about development of this module.

AUTHOR

Karl Erisman (karl.erisman@icainformatics.com)