I hesitate to answer as this is my first post to PM, but I've had similar problems in the past and think I can help. In an effort to avoid hand-coding SQL statements, I've opted for the following:

package Node;
use base 'My::Class::DBI';

... # standard Class::DBI stuff here

# Declare that we have a relationships table that
# tells us where our children are
Node->has_many('children_refs' => 'Relationship' => 'parent_id');

# Now that we can fetch those records from the
# relationships table, we need to have a way to
# get Node objects from those relationships
sub children {
  return Node->search_in(
    Node->primary_column => [
      map { $_->child_id } shift->children_refs
    ]
  );
}

# Similarly for parents
[download]

Now you'll need that search_in method as well.

package My::Class::DBI;
use base 'Class::DBI';

sub search_in {
  my $class = shift;
  my %args = ref $_[0] eq 'HASH' ? ${ $_[0] } : @_;

  # Normalize columns
  my @columns = keys %args;
  $_ = $class->_normalized($_) for @columns;

  # Check columns
  $class->_check_columns(@columns);

  # Keep list of all values (for each column)
  # that'll be used in the IN clause -- we'll
  # later bind these values to an SQL statement
  # so that the DBI can take care of the quoting
  my @values;
  my $sql = join ' AND ', map {
    # Grab values for IN clause
    my @vals = @{ $args{$_} };
    push @values, @vals;
    
    my $in_str = join(',', ('?') x @vals);
    "$_ IN ($in_str)" if @vals;
  } @columns;
 
  return unless @values;
  return $class->retrieve_from_sql($sql, @values);
}
[download]

With the exception of the search_in method, this just might be a nice and tidy solution. I especially prefer this strategy as it doesn't rely on hand-coded SQL satements.

Good luck!
David