Let me expand on what arOn was saying

Suppose you ultimately want to call foo->bar( $baz ). And the way you want to achieve it is by dynamically deciding what the subroutine you're going to use, like so:

    my $subname = 'baz';
    my $param   = { abc => 'xyz' }; # or an object. whatever.

    eval 'foo->' . "$subname( \$param )";
[download]

Okay, now observe what happens when you call the eval.

    # first, the string that is to be eval'ed is created
    'foo->' . "$subname( \$param )"  => 'foo->baz( $param )'

    # then the eval kicks in
    eval 'foo->baz( $param )'

    # which is effectively means
    foo->baz( $param );

    # where $param is the ref to a hash
[download]

Now observe what happens if I change the quotations and such...

    # suppose we use "$subname( $param )"
    'foo->' . "$subname( $param )" => "foo->baz( 'HASHx(....)' )"

    # where HASHx(...) is the string representation of the
    # hash ref. this happens because you're interpolating
    # the hashref within the string
    end result => foo->baz( 'HASHx( .... )' );

----
    # what if we use "baz( $param )" ?
    'foo->' . "baz( $param )" => "foo->baz( 'HASHx(....)' )"

    # ah, same as the example above
    end result => foo->baz( 'HASHx( .... )' );

----
    # suppose we use '$subname( $param )'
    'foo->' . '$subname( $param )' => 'foo->$subname( $param )'

    # so no interpolation is done in the first step...
    foo->$subname( $param ); # ah, but this is perfectly valid!
[download]

Hope that clears things up. Personally, I like suaveant's way of calling method better, though