Check out the section on the has_many and has_a methods. You will want to add something like the following to your classes:

# To your authors class
__PACKAGE__->has_many('books', 'My::DB::Books' => book_id);

# To your books class
__PACKAGE__->has_a('author_id', 'My::DB::Authors');
[download]

Then in your code you can use:

my $author = My::DB::Authors->retrieve($author_id);

# method add_to_books is created for you by has_many
$author->add_to_books({
                       col1 => value1,
                       col2 => value2,
                       ...
});  # author_id will be filled in automatically

my $books = $author->books;
while (my $book = $books->next) {
  ...
}
[download]

- Cees

Not quite. This would be correct if there were a one-to-many relationship between authors and books (ie if each book was allowed only one author), but the OP has a many-to-many relationship there - presumably on the grounds that some books have several authors - and, like everyone, is wondering how to make the joining table invisible. The short answer is that s/he does have to create a My::DB::AuthorBook joining class, but s/he will probably never need to access it directly. It's in the pod under 'mapping', a subheading of 'relationships'.

the old way of doing this was to set up an AuthorBook class that had_a book and had_a author, and then use the nearly ubiquitous idiom:

Package Author;
Author->has_many('_book_links', 'AuthorBook', 'author');
sub books { return map { $_->book } shift->_book_links; }
[download]

but as of 0.91ish, you can achieve exactly the same thing with:

Package Author;
Author->has_many('books', [ AuthorBook => 'book' ], 'author');
[download]

The syntax is getting rather arbitrary (I'd really like to see named parameters there), and you still have to create the nearly-empty AuthorBook class, but that's kind of a good thing - there will probably be methods you want to add to it one day - and at least this way the relationships are defined where it makes most sense to look for them.