=head1 NAME

Tangram::Type::Array::FromOne - map Perl arrays using a foreign key

=head1 SYNOPSIS

   use Tangram;

   # or
   use Tangram::Core;
   use Tangram::Type::Array::FromOne;

   $schema = Tangram::Schema->new(

      classes => { Agenda => { fields => {

      iarray =>
      {
         # long form
         entries =>
         {
            class => 'Entry',
            coll => 'agenda',
         },

         # or (short form)
         entries => 'Entry',
      }

=head1 DESCRIPTION

This class maps references to a Perl array in an intrusive
fashion. The persistent fields are grouped in a hash under the
C<iarray> key in the field hash.

The array may contain only objects of persistent classes. These classes
must have a common persistent base class.

Tangram uses two columns on the element's table to store:

=over 4

=item * the id of the object containing the collection

=item * the position of the element in the collection

=back

CAUTION: the same object may not be an element of the same collection,
in two different objects. This mapping may be used only for
one-to-many relationships.

The field names are passed in a hash that associates a field name with
a field descriptor. The field descriptor may be either a hash or a
string. The hash uses the following fields:

=over 4

=item * class

=item * aggreg

=item * back

=item * coll

=item * slot

=item * deep_update

=back

Mandatory field C<class> specifies the class of the elements.

Optional field C<aggreg> specifies that the elements of the collection
must be removed (erased) from persistent storage along with the
containing object. The default is not to aggregate.

Optional field C<back> sets the name of a field that is inserted in
the elements. That field acts as a demand-loaded, read-only reference
to the object containing the collection.

Optional field C<coll> sets the name the column containing the id of
the containing object. This defaults to 'C_m', where 'C' is the class
of the containing object, and 'm' is the field name.

Optional field C<slot> sets the name the column containing the id of
the containing object. This defaults to 'C_m_slot', where 'C' is the
class of the containing object, and 'm' is the field name.

The "C" in C_m and C_m_slot are passed through the schema
normalisation function before being combined into a column name.

Optional field C<deep_update> specificies that all elements have to be
updated automatically when C<update> is called on the collection
object. Automatic update ensures consisitency between the Perl
representation and the DBMS state, but degrades update performance so
use it with caution. The default is not to do automatic updates.

If the descriptor is a string, it is interpreted as the name of the
element's class. This is equivalent to specifying only the C<class>
field in the hash variant.