DBIx::Class 0.08001 released (and a podcast)

For those of you who're interested in me demonstrating just how badly I interview, you can hear a perlcast podcast interview with me about DBIx::Class which was recorded last month.

Better still, after 7 dev releases, four of them RC grade, a lot of bug fixing, a lot of hating databases, and sterling work from many many people, I'm proud to announce that DBIx::Class 0.08000 is on its way to being available from the CPAN! ... and of course two minutes after I wrote this, somebody found a bug. So please, please make sure you get 0.08001, not 0.08000.

Given it's been not that far short of a year since the last major release, there are quite a lot of new features to show for the time we've spent, so I'm just going to mention a few highlights here.

First, a short list of feature/bug stuff:

• deep prefetch now works for a lot more cases (and is completely rewritten to be clearer and faster)
• inflate_column now obeys 'accessor' column_info key for accessor naming
• multiple ordering columns in DBIx::Class::Ordered
• support for BYTEA and similar columns in Pg (and general support so other drivers can be updated)
• ANSI join syntax now unrolled into WHERE clause to support Oracle 8 and below
• modified reference approaches so $schema objects don't go out of scope anymore
• more efficient database disconnect handling and cleaner error reporting
• significant speed improvements in common row operations
• experimental Replication storage class for "modify master, select from slave" setups

There are also a few bigger things that I'd like to highlight.

DBIx::Class::Schema->load_namespaces

This is an alternative to load_classes which assumes an alternative layout for automatic class loading.  It assumes that all result classes are underneath a sub-namespace of the schema called Result, any
corresponding ResultSet classes are underneath a sub-namespace of the schema
called ResultSet.

For example:

  # load My::Schema::Result::CD, My::Schema::Result::Artist,
  #    My::Schema::ResultSet::CD, etc...
  My::Schema->load_namespaces;

  # Override everything to use ugly names.
  # In this example, if there is a My::Schema::Res::Foo, but no matching
  #   My::Schema::RSets::Foo, then Foo will have its
  #   resultset_class set to My::Schema::RSetBase
  My::Schema->load_namespaces(
    result_namespace => 'Res',
    resultset_namespace => 'RSets',
    default_resultset_class => 'RSetBase',
  );

  # Put things in other namespaces
  My::Schema->load_namespaces(
    result_namespace => '+Some::Place::Results',
    resultset_namespace => '+Another::Place::RSets',
  );

DBIx::Class::Row->new/insert/update

DBIx::Class will now correctly handle being passed a hashref (for a one-one or many-one relationship) or arrayref of hashrefs (for a one-many relationship) at new() or update() time and the correct data will get created - for example

my $artist = $schema->resultset('Artist')
                                   ->create({ name => 'Fred 2',
                                                     cds => [
                                                             { title => 'Music to code by',
                                                               year => 2007,
                                                             },
                                                             ],
                                                     });

# Add a new CD
$artist->update({cds => [ $artist->cds,
                          { title => 'Yet another CD',
                            year => 2006,
                          },
                        ],
                });

DBIx::Class also now has experimental schema versioning support. Adding

  __PACKAGE__->load_components(qw/+DBIx::Class::Schema::Versioned/);
  __PACKAGE__->upgrade_directory('/path/to/upgrades/');
  __PACKAGE__->backup_directory('/path/to/backups/');

to your schema class, and calling $schema->create_ddl_dir to generate a directory of SQL files for your current schema version (which is best practice since it then allows deploy to work on target systems without SQL::Translator present) will allow you to call $schema->upgrade to do upgrades. See the perldoc for more details.

Have fun!