++ed by:

1 non-PAUSE user.

Doug Hoyte


Object::Sub - Create objects without those pesky classes


    use Object::Sub;

    my $obj = Object::Sub->new(sub {
                  my ($self, $method, @args) = @_;

                  print "self: $self, method name: $method, first arg: $args[0]\n";

    ## self: Object::Sub=HASH(0xc78eb0), method name: whatever, first arg: 123

    ## self: Object::Sub=HASH(0xc78eb0), method name: , first arg: 123
    ##   ($method is undef)

Alternatively, you can use a hash of subs:

    my $obj = Object::Sub->new({
        add => sub {
            my ($self, $num1, $num2) = @_;
            return $num1 + $num2;
        mul => sub {
            my ($self, $num1, $num2) = @_;
            return $num1 * $num2;

    $obj->add(2, 3);
    ## => 5


Sometimes you want something that acts like an object but you don't want to go to all the trouble of creating a new package, with constructor and methods and so on. This module is a trivial wrapper around perl's AUTOLOAD functionality which intercepts method calls and lets you handle them in a single sub. It also uses overload so that you can additionally treat the object as a sub if you desire.



AUTOLOAD allows you to dispatch on method names at run-time which can sometimes be useful, for example in RPC protocols where you transmit method call messages to another process for them to be executed remotely. Unfortunately, using AUTOLOAD is a bit annoying since the interface is somewhat arcane. Object::Sub is a nicer interface to the most commonly used AUTOLOAD functionality:

    my $obj = Object::Sub->new(sub {
                my ($self, $method, @args) = @_;

                my $rpc_input = encode_json({ method => $method, args => [ @args ] });

                my $rpc_output = do_rpc_call($rpc_input);

                return decode_json($rpc_output);

Because Object::Sub objects can also be treated as subs, your RPC interface can support sub-routine calls on the objects as well as method calls, even on the same object.


Some APIs require you to pass in or provide an object but then don't actually end up using it. Instead of passing in undef and getting a weird Can't call method "XYZ" on an undefined value error, you can pass in an Object::Sub which will throw a "helpful" exception instead:

    my $obj = Some::API->new(
                logger => Object::Sub->new(sub { die "FIXME: add logger" }),

Alternatively, you may choose to minimally implement the API "inline" in your program:

    my $obj = Some::API->new(
                logger => Object::Sub->new(sub {
                            my ($self, $method, @args) = @_;

                            return if $method eq 'debug';

                            say STDERR "Some::API $method: " . join(' ', @args);


Again, some APIs may never end up using an object so you may wish to "lazily" defer the creation of that object until a method is actually called on it. This module can help you make the cases where it doesn't use it more efficient.

For example, suppose you have a large CGI script which always opens a DBI connection but only actually accesses this connection for a small portion of runs. You can prevent the script from accessing the database on the majority of runs with Object::Sub:

    my $dbh = Object::Sub->new(sub {
                require DBI;
                $_[0] = DBI->connect($dsn, $user, $pass, { RaiseError => 1 })
                    || die "Unable to connect to database: $DBI::errstr";

                my ($self, $method, @args) = @_;
                return $self->$method(@args);

Note how we don't even load or compile the module until the first method is called. After you call a method on $dbh it changes from a Object::Sub object into a DBI object (assuming the DBI->connect constructor succeeds). This works because the $_[0] argument is actually an alias to $dbh and can be modified.

To demonstrate this, here is an example with Session::Token:

    my $o = Object::Sub->new(sub {
              require Session::Token;
              $_[0] = Session::Token->new;

              my ($self, $method, @args) = @_;
              return $self->$method(@args);

    say ref $o;
    ## Object::Sub

    say $o->get;
    ## mhDPtfLlFMGl5kyNcJgFt7

    say ref $o;
    ## Session::Token

    say $o->get;
    ## 4JYkGgwWbYWGleU7Qk912P

With Object::Sub you can lazily "create" and pass around objects before their constructor code has even been loaded.


Although not really a bug in this module, common perl code tends to copy references of objects. Any code that overwrites the caller object (for example in the "LAZY OBJECT CREATION" section) will only update one of the copies.


Object-Sub github repo


Doug Hoyte, <doug@hcsw.org>


Copyright 2015-2016 Doug Hoyte.

This module is licensed under the same terms as perl itself.