kudra has asked for the wisdom of the Perl Monks concerning the following question:

I've got a question which I hope someone can answer. The problem has been solved, but I'm curious about why the solution was neccessary.

Note: All examples are pseudocode.

I have been working on an application which runs in the fast CGI environment. Much of the functionality has been put in modules. There are a number of variables which need to be shared by the modules in a single program run, but which must be distinct each time the application is run. One of these variables is a DBI handler, which I was passing to the constructors of two different modules:

my $one = new One(dbh => $dbh, ...); my $two = new Two(dbh => $dbh, ...);
Module One parses some files, begins a transaction, and calls routines in module Two to insert some data. The commit is done in module One. The handle is disconnected in the destruction of Two.

This did not work. It was almost as if One and Two had different connections, because when I added a commit to Two, the data was committed. Yet I did not get any warnings about the handle in Two being destroyed instead of disconnected. In short, it didn't act like a new handler, nor did it act like an old one.

I tried again to make sure that One and Two got the same handler. I changed the code to:

my $one = new One(dbh => \$dbh, ...); my $two = new Two(dbh => \$dbh, ...);
This worked.

My question is: why did I need to explicitly pass this by reference?