Question

When dealing with persistant MySQL connections, the one problem is that they get dropped after a certain timeout (usually 28800 seconds).
DBIx::Connector seems to do the job of automatically reconnecting to a dropped connection, but it adds more Perl code to each SQL statement which can get annoying. For example instead of:

    $dbh->do('DROP DATABASE stackoverflow');

One has to say:

    $conn->run(
        fixup => sub {
            my $dbh = shift;
            $dbh->do('DROP DATABASE stackoverflow');
        }
    );

Suppose one does not need transactions. Why would one want to use DBIx::Connector instead of passing $dbh->{mysql_auto_reconnect} = 1, which also works well?

Answer 1

1. DBIx::Connector‘s stated goal is to provide a fork- and thread-safe implementation of DBI‘s connect_cached(). So you’re almost asking an apples/oranges question.

   However, DBIx::Connector does also reconnect if the connection is lost, when it is running in either its ping or fixup Connection Modes. Note that the default is the no_ping mode, which apparently does not attempt reconnection.

2. DBIx::Connector will work with any DB backend, not just MySQL.

All said… if you’re using MySQL, and don’t care about the other advantages of DBIx::Connector (because you never fork or use threads, for instance), then mysql_auto_reconnect is probably perfect for you.