turn off cascade_delete/copy, set on_delete/update

[dbsrgits/DBIx-Class-Schema-Loader.git] / lib / DBIx / Class / Schema / Loader / Manual / UpgradingFromV4.pod

diff --git a/lib/DBIx/Class/Schema/Loader/Manual/UpgradingFromV4.pod b/lib/DBIx/Class/Schema/Loader/Manual/UpgradingFromV4.pod
index c5869aa..21dc9e3 100644 (file)
--- a/lib/DBIx/Class/Schema/Loader/Manual/UpgradingFromV4.pod
+++ b/lib/DBIx/Class/Schema/Loader/Manual/UpgradingFromV4.pod
@@ -20,13 +20,19 @@ C<is_auto_increment> that it didn't set before.
 
 RelBuilder
 
-The new RelBuilder will give you much nicer accessor names for relationships,
-so you will no longer have conflicts between a foreign key column and the
-relationship accessor itself.
+The new RelBuilder will give you nicer accessor names for relationships, so you
+will no longer have conflicts between a foreign key column and the relationship
+accessor itself (if the FK is named C<_id>.)
 
 It will also more correctly infer the relationship type, e.g. some relationships
 that were previously detected as a C<has_many> will now be a C<might_have>
-(when the foreign key refers to a unique index.)
+(when it detects a unique constraint on the foreign key column.)
+
+Also C<cascade_delete> and C<cascade_copy> are turned off for by default for
+C<has_many> and C<might_have> relationships, while C<belongs_to> relationships
+are created with C<< on_delete => 'CASCADE' >> and C<< on_update => 'CASCADE' >>
+by default. This is overridable via
+L<relationship_attrs|DBIx::Class::Schema::Loader::Base/relationship_attrs>.
 
 =item *
 
@@ -38,34 +44,45 @@ becomes C<UserRole> instead.
 
 =item *
 
+use_namespaces
+
+Now defaults to on. See L<DBIx::Class::Schema::Loader::Base/use_namespaces> and
+L<DBIx::Class::Schema/"load_namespaces">.
+
+=item *
+
 Support for more databases
 
-We now support Microsoft SQL Server and Sybase, and there are also improvements
-to the other backends.
+We now support Microsoft SQL Server and Sybase, and there are also many
+improvements to the other backends.
 
 =back
 
 =head1 Backward Compatibility
 
 In backward compatibility mode, the Loader will use the old relationship names
-and types, and will not singularize monikers for tables.
+and types, will not singularize monikers for tables, and C<use_namespaces> will
+be off.
 
-To control this behavior see L<DBIx::Class::Schema::Loader::Base/naming>.
+To control this behavior see L<DBIx::Class::Schema::Loader::Base/naming> and
+L<DBIx::Class::Schema::Loader::Base/use_namespaces>.
 
 =head2 Static Schemas
 
 When reading a C<Schema.pm> from a static schema generated with an C<0.04>
-version of Loader, backward compatibility mode willl be turned on, unless
-overridden with the naming accessor.
+version of Loader, backward compatibility mode will default to on, unless
+overridden with the C<naming> and/or C<use_namespaces> attributes.
 
 =head2 Dynamic Schemas
 
-Dynamic schemas will always by default use C<0.04006> mode.
+Dynamic schemas will always by default use C<0.04006> mode and have
+C<use_namespaces> off.
 
-To upgrade a dynamic schema, set the naming accessor (which is proxied to the
-loader) in your C<Schema.pm>:
+To upgrade a dynamic schema, set the C<naming> and C<use_namespaces> attributes
+(which is proxied to the loader) in your C<Schema.pm>:
 
     __PACKAGE__->naming('current');
+    __PACKAGE__->use_namespaces(1);
 
 =head1 AUTHOR