"Class->install_properties(\%params)"¶

Defines all the properties of the specified object class. Generally you should call install_properties() in the body of your class definition, so the properties can be set when the class is "use"d or "require"d.

Required members of %params are:

• "columns"

  All the columns in the object class. This property is an arrayref.

• "datasource"

  The identifier of the table in which the object class's data are stored. Usually the datasource is simply the table name, but the datasource can be decorated into the table name by the "Data::ObjectDriver::DBD" module if the database requires special formatting of table names.

• "driver" or "get_driver"

  The driver used to perform database operations (lookup, update, etc) for the object class.

  "driver" is the instance of "Data::ObjectDriver" to use. If your driver requires configuration options not available when the properties are initially set, specify a coderef as "get_driver" instead. It will be called the first time the driver is needed, storing the driver in the class's "driver" property for subsequent calls.

The optional members of %params are:

• "primary_key"

  The column or columns used to uniquely identify an instance of the object class. If one column (such as a simple numeric ID) identifies the class, "primary_key" should be a scalar. Otherwise, "primary_key" is an arrayref.

• "column_defs"

  Specifies types for specially typed columns, if any, as a hashref. For example, if a column holds a timestamp, name it in "column_defs" as a "date" for proper handling with some "Data::ObjectDriver::Driver::DBD" database drivers. Columns for which types aren't specified are handled as "char" columns.

  Known "column_defs" types are:

•

"db"

The name of the database. When used with "Data::ObjectDriver::Driver::DBI" type object drivers, this name is passed to the "init_db" method when the actual database handle is being created.

Custom object drivers may define other properties for your object classes. Consult the documentation of those object drivers for more information.

"Class->install_column($col, $def)"¶

Modify the Class definition to declare a new column $col of definition <$def> (see column_defs).

"Class->has_a(@definitions)"¶

NOTE: "has_a" is an experimental system, likely to both be buggy and change in future versions.

Defines a foreign key reference between two classes, creating accessor methods to retrieve objects both ways across the reference. For each defined reference, two methods are created: one for objects of class "Class" to load the objects they reference, and one for objects of the referenced class to load the set of "Class" objects that reference them.

For example, this definition:

    package Ingredient;
    __PACKAGE__->has_a(
        { class => 'Recipe', column => 'recipe_id' },
    );

would create "Ingredient->recipe_obj" and "Recipe->ingredient_objs" instance methods.

Each member of @definitions is a hashref containing the parameters for creating one accessor method. The required members of these hashes are:

• "class"

  The class to associate.

• "column"

  The column or columns in this class that identify the primary key of the associated object. As with primary keys, use a single scalar string for a single column or an arrayref for a composite key.

The optional members of has_a() definitions are:

• "method"

  The name of the accessor method to create.

  By default, the method name is the concatenated set of column names with each "_id" suffix removed, and the suffix "_obj" appended at the end of the method name. For example, if "column" were "['recipe_id', 'ingredient_id']", the resulting method would be called "recipe_ingredient_obj" by default.

• "cached"

  Whether to keep a reference to the foreign object once it's loaded. Subsequent calls to the accessor method would return that reference immediately.

• "parent_method"

  The name of the reciprocal method created in the referenced class named in "class".

  By default, that method is named with the lowercased name of the current class with the suffix "_objs". For example, if in your "Ingredient" class you defined a relationship with "Recipe" on the column "recipe_id", this would create a "$recipe->ingredient_objs" method.

  Note that if you reference one class with multiple sets of fields, you can omit only one parent_method; otherwise the methods would be named the same thing. For instance, if you had a "Friend" class with two references to "User" objects in its "user_id" and "friend_id" columns, one of them would need a "parent_method".

"Class->has_partitions(%param)"¶

Defines that the given class is partitioned, configuring it for use with the "Data::ObjectDriver::Driver::SimplePartition" object driver. Required members of %param are:

• "number"

  The number of partitions in which objects of this class may be stored.

• "get_driver"

  A function that returns an object driver, given a partition ID and any extra parameters specified when the class's "Data::ObjectDriver::Driver::SimplePartition" was instantiated.

Note that only the parent object for use with the "SimplePartition" driver should use has_partitions(). See "Data::ObjectDriver::Driver::SimplePartition" for more about partitioning.

"Class->lookup($id)"¶

Returns the instance of "Class" with the given value for its primary key. If "Class" has a complex primary key (more than one column), $id should be an arrayref specifying the column values in the same order as specified in the "primary_key" property.

"Class->search(\%terms, [\%args])"¶

Returns all instances of "Class" that match the values specified in "\%terms", keyed on column names. In list context, "search" returns the objects containing those values. In scalar context, "search" returns an iterator function containing the same set of objects.

Your search can be customized with parameters specified in "\%args". Commonly recognized parameters (those implemented by the standard "Data::ObjectDriver" object drivers) are:

• "sort"

  A column by which to order the object results.

• "direction"

  If set to "descend", the results (ordered by the "sort" column) are returned in descending order. Otherwise, results will be in ascending order.

• "limit"

  The number of results to return, at most. You can use this with "offset" to paginate your search() results.

• "offset"

  The number of results to skip before the first returned result. Use this with "limit" to paginate your search() results.

• "fetchonly"

  A list (arrayref) of columns that should be requested. If specified, only the specified columns of the resulting objects are guaranteed to be set to the correct values.

  Note that any caching object drivers you use may opt to ignore "fetchonly" instructions, or decline to cache objects queried with "fetchonly".

• "for_update"

  If true, instructs the object driver to indicate the query is a search, but the application may want to update the data after. That is, the generated SQL "SELECT" query will include a "FOR UPDATE" clause.

All options are passed to the object driver, so your driver may support additional options.

"Class->result(\%terms, [\%args])"¶

Takes the same %terms and %args arguments that search takes, but instead of executing the query immediately, returns a Data::ObjectDriver::ResultSet object representing the set of results.

"$obj->exists()"¶

Returns true if $obj already exists in the database.

"$obj->save()"¶

Saves $obj to the database, whether it is already there or not. That is, save() is functionally:

    $obj->exists() ? $obj->update() : $obj->insert()

"$obj->update()"¶

Saves changes to $obj, an object that already exists in its database.

"$obj->insert()"¶

Adds $obj to the database in which it should exist, according to its object driver and configuration.

"$obj->remove()"¶

Deletes $obj from its database.

"$obj->replace()"¶

Replaces $obj in the database. Does the right thing if the driver knows how to REPLACE object, ala MySQL.

"Class->new(%columns)"¶

Returns a new object of the given class, initializing its columns with the values in %columns.

"$obj->init(%columns)"¶

Initializes $obji by initializing its columns with the values in %columns.

Override this method if you must do initial configuration to new instances of $obj's class that are not more appropriate as a "post_load" callback.

"Class->properties()"¶

Returns the named object class's properties as a hashref. Note that some of the standard object class properties, such as "primary_key", have more convenient accessors than reading the properties directly.

"Class->driver()"¶

Returns the object driver for this class, invoking the class's get_driver function (and caching the result for future calls) if necessary.

"Class->get_driver($get_driver_fn)"¶

Sets the function used to find the object driver for Class objects (that is, the "get_driver" property).

Note that once driver() has been called, the "get_driver" function is not used. Usually you would specify your function as the "get_driver" parameter to install_properties().

"Class->is_pkless()"¶

Returns whether the given object class has a primary key defined.

"Class->is_primary_key($column)"¶

Returns whether the given column is or is part of the primary key for "Class" objects.

"$obj->primary_key()"¶

Returns the values of the primary key fields of $obj.

"Class->primary_key_tuple()"¶

Returns the names of the primary key fields of "Class" objects.

"$obj->is_same($other_obj)"¶

Do a primary key check on $obj and $<other_obj> and returns true only if they are identical.

"$obj->object_is_stored()"¶

Returns true if the object hasn't been stored in the database yet. This is particularly useful in triggers where you can then determine if the object is being INSERTED or just UPDATED.

"$obj->pk_str()"¶

returns the primary key has a printable string.

"$obj->has_primary_key()"¶

Returns whether the given object has values for all of its primary key fields.

"$obj->uncache_object()"¶

If you use a Cache driver, returned object will be automatically cached as a result of common retrieve operations. In some rare cases you may want the cache to be cleared explicitly, and this method provides you with a way to do it.

"$obj->primary_key_to_terms([$id])"¶

Returns $obj's primary key as a hashref of values keyed on column names, suitable for passing as search() terms. If $id is specified, convert that primary key instead of $obj's.

"Class->datasource()"¶

Returns the datasource for objects of class "Class". That is, returns the "datasource" property of "Class".

"Class->columns_of_type($type)"¶

Returns the list of columns in "Class" objects that hold data of type $type, as an arrayref. Columns are of a certain type when they are set that way in "Class"'s "column_defs" property.

"$obj->set_values(\%values)"¶

Sets all the columns of $obj that are members of "\%values" to the values specified there.

"$obj->set_values_internal(\%values)"¶

Sets new specified values of $obj, without using any overridden mutator methods of $obj and without marking the changed columns changed.

"$obj->clone()"¶

Returns a new object of the same class as $obj containing the same data, except for primary keys, which are set to "undef".

"$obj->clone_all()"¶

Returns a new object of the same class as $obj containing the same data, including all key fields.

"Class->has_column($column)"¶

Returns whether a column named $column exists in objects of class <Class>.

"Class->column_names()"¶

Returns the list of columns in "Class" objects as an arrayref.

"$obj->column_values()"¶

Returns the columns and values in the given object as a hashref.

"$obj->column($column, [$value])"¶

Returns the value of $obj's column $column. If $value is specified, column() sets the first.

Note the usual way of accessing and mutating column values is through the named accessors:

    $obj->column('fred', 'barney');  # possible
    $obj->fred('barney');            # preferred

"$obj->is_changed([$column])"¶

Returns whether any values in $obj have changed. If $column is given, returns specifically whether that column has changed.

"$obj->changed_cols_and_pk()"¶

Returns the list of all columns that have changed in $obj since it was last loaded from or saved to the database, as a list.

"$obj->changed_cols()"¶

Returns the list of changed columns in $obj as a list, except for any columns in $obj's primary key (even if they have changed).

"Class->lookup_multi(\@ids)"¶

Returns a list (arrayref) of objects as specified by their primary keys.

"Class->bulk_insert(\@columns, \@data)"¶

Adds the given data, an arrayref of arrayrefs containing column values in the order of column names given in "\@columns", as directly to the database as "Class" records.

Note that only some database drivers (for example, "Data::ObjectDriver::Driver::DBD::Pg") implement the bulk insert operation.

"$obj->fetch_data()"¶

Returns the current values from $obj as saved in the database, as a hashref.

"$obj->refresh()"¶

Resets the values of $obj from the database. Any unsaved modifications to $obj will be lost, and any made meanwhile will be reflected in $obj afterward.

"$obj->column_func($column)"¶

Creates an accessor/mutator method for column $column, returning it as a coderef.

Override this if you need special behavior in all accessor/mutator methods.

"$obj->deflate()"¶

Returns a minimal representation of the object, for use in caches where you might want to preserve space (like memcached). Can also be overridden by subclasses to store the optimal representation of an object in the cache. For example, if you have metadata attached to an object, you might want to store that in the cache, as well.

"Class->inflate($deflated)"¶

Inflates the deflated representation of the object $deflated into a proper object in the class Class. That is, undoes the operation "$deflated = $obj->deflate()" by returning a new object equivalent to $obj.

Introduction¶

When dealing with the methods on this class, the transactions are global, i.e: applied to all drivers. You can still enable transactions per driver if you directly use the driver API.

"Class->begin_work"¶

This enable transactions globally for all drivers until the next rollback or commit call on the class.

If begin_work is called while a transaction is still active (nested transaction) then the two transactions are merged. So inner transactions are ignored and a warning will be emitted.

"Class->rollback"¶

This rollbacks all the transactions since the last begin work, and exits from the active transaction state.

"Class->commit"¶

Commits the transactions, and exits from the active transaction state.

"Class->txn_debug"¶

Just return the value of the global flag and the current working drivers in a hashref.

"Class->txn_active"¶

Returns true if a transaction is already active.