I usually just write a little MyDB module that has one method: connect. Then any script that wants to use the database just does:

use MyDB;
my $dbh = MyDB::connect();
[download]

This method will also let you add connection pooling later without changing any code other than the MyDB module.

Here's a sample MyDB module with a connect method that lets you override any of the default connect parameters (note: I just rewrote a good chunk of this as I posted it, and haven't tested it, so...):

package MyDB;

use DBI;

my %Default =
    ( database => 'mydatabase',
      driver => 'mysql',
      hostname => 'localhost',
      port => 3306,
      username => 'someuser',
      password => 'topsecret',
      options => { AutoCommit => 1,
                   RaiseError => 0,
                 }
      );

sub connect {
    my %params = @_;
    my $connect_str=sprintf("dbi:%s:database=%s;hostname=%s;port=%s",
        $params{driver} || $Default{driver},
        $params{database} || $Default{database},
        $params{hostname} || $Default{hostname},
        $params{port} || $Default{port});
        
    return DBI->connect($connect_str,
                        $params{username} || $Default{username},
                        $params{password} || $Default{password},
                        $params{options} || $Default{options}
                     );
}

1;
[download]

Brad