Badger::Hub
- NAME
- SYNOPSIS
- INTRODUCTION
- DESCRIPTION
- Components
- Delegates
- Subclassing Badger::Hub
- Circular References are a Good Thing
- METHODS
- new()
- components()
- component($name)
- delegates()
- delegate($name)
- destroy()
- INTERNAL METHODS
- construct($component,\%params)
- auto_can($name)
- auto_component($name,$module)
- auto_delegate($name,$component)
- config()
- AUTHOR
- COPYRIGHT
This documentation describes the Badger::Hub
object. A hub
sits in the middle of a Badger
application, providing a central point of access to the various other
modules, components and sub-system that an application uses.
You generally don't need to worry about the Badger::Hub
if
you're just a casual user of the Badger modules. It will primarily be of interest to developers
who are building their own badger-powered applications or extensions.
At present this module is quite basic. It will be developed further in due course.
A Badger::Hub
object is a central repository of shared
resources for a Badger
application. The hub sits in the middle of an application and provides
access to all the individual components and larger sub-systems that may
be required. It automatically loads and instantiates these other modules
on demand and caches then for subsequent use.
The Badger::Hub base class currently has two components:
filesystem => Badger::Filesystem codecs => Badger::Codecs
An AUTOLOAD
method allows you to access any component by
name. It will be loaded and instantiated automatically. The
AUTOLOAD
method also generates the missing method so that
you can avoid the overhead of the AUTOLOAD
method the next
time you call it.
my $filesystem = $hub->filesystem;
You can add your own component to a hub and they will be available in the same way.
$hub->components( fuzzbox => 'My::Module::Fuzzbox' ); my $fuzzbox = $hub->fuzzbox;
As well as accessing components directly, you can also make use of
delegate methods that get forwarded onto a component. For example, the
hub file()
method is just a short cut to the
file()
method of the filesystem
component
(implemented by Badger::Filesystem).
$file = $hub->file('/path/to/file'); # the short cut $file = $hub->filesystem->file('/path/to/file'); # the long way
You can easily define your own delegate methods.
$hub->delegates( warm_fuzz => 'fuzzbox' ); $fuzzed = $hub->warm_fuzz; # the short way $fuzzed = $hub->fuzzbox->warm_fuzz; # the long way.
You can subclass Badger::Hub to define your own collection of components and delegate methods, as shown in the example below.
package My::Hub; use Badger::Class version => 0.01, debug => 0, base => 'Badger::Hub'; our $COMPONENTS = { fuzzbox => 'My::Module::Fuzzbox', flanger => 'My::Module::Flanger', }; our $DELEGATES = { warm_fuzz => 'fuzzbox', dirty_noise => 'fuzzbox', wide_flange => 'flanger', wet_flange => 'flanger', };
In some cases, sub-systems instantiated by a Badger::Hub will also maintain a reference back to the hub. This allows them to access other sub-systems and components that they require.
Note that this behaviour implicitly creates circular references between the hub and its delegates. This is intentional. It ensures that the hub and delegates keep each other alive until the hub is explicitly destroyed and the references are freed. Having the hub stick around for as long as possible is usually a Good Thing. It acts as a singleton providing a central point of access to the resources that your application uses (which is a fancy way of saying it's like a global variable).
+-----+ +-----------+ | HUB |----->| COMPONENT | | |<-----| | +-----+ +-----------+
If you manually create a hub for whatever reason (and the cases where you would need to are few and far between) then you are responsible for calling the destroy() method when you're done with it. This will manually break the circular references and free up any memory used by the hub and any delegates it is using. If you don't call the destroy() method then the hub will remain alive until the end of the program when the memory will be freed as usual. In most cases this is perfectly acceptable.
However, you generally don't need to worry about any of this because you
wouldn't normally create a hub manually. Instead, you would leave it up
to the Badger façade (or
"front-end") module to do that behind the scenes. When you create
a Badger module it implicitly
creates a Badger::Hub
to use. When the Badger object goes out of scope its
DESTROY
method automatically calls the hub's destroy method.
sub foo { my $badger = Badger->new; my $hub = $badger->hub; # do something # $badger object is freed here, that calls $hub->destroy }
Because there is no reference from the hub back to the Badger façade object you don't have to worry about circular references. The Badger object is correctly freed and that ensures the hub gets cleaned up.
+--------+ +-----+ +-----------+ | BADGER |----->| HUB |----->| COMPONENT | | | | |<-----| | +--------+ +-----+ +-----------+
If you call Badger
methods as class methods then they are
forwarded to a prototype object (effectively a singleton object). That in turn
will use a prototype
hub object. In this case, both the Badger
and
Badger::Hub
objects will exist until the end of the program.
This ensures that your class methods all Do the right Thing
without you having to worry about creating a Badger object.
# class method creates Badger prototype, which creates Badger::Hub # prototype, which loads, instantiates and caches Badger::Filesystem # which can then fetch the file my $file = Badger->file('/path/to/file'); # later... reuse same Badger, Badger::Hub and Badger::Filesystem my $dir = Badger->dir('/path/to/dir');
This method can be used to get or set entries in the components table for the hub. Components are other modules that the hub can delegate to.
# get components hash ref my $comps = $hub->components; # add new components $hub->components({ fuzzbox => 'My::Module::Fuzzbox', flanger => 'My::Module::Flanger', });
This method returns a single entry from the components table.
print $hub->component('fuzzbox'); # My::Module::Fuzzbox
This method can be used to get or set entries in the delegates table for the hub. This specifies which hub methods should be delegated to components.
# get delegates hash ref my $delegs = $hub->delegates; # add new delegates $hub->delegates({ warm_fuzz => 'fuzzbox', dirty_noise => 'fuzzbox', wide_flange => 'flanger', wet_flange => 'flanger', });
This method returns a single entry from the delegates table.
print $hub->delegate('warm_fuzz'); # fuzzbox
This method configures and instantiates a component. The first argument
is the component name. This is mapped to a module via the component() method and the module is loaded.
A list of named parameters, or a reference to a hash array of named
paramters may follow. A reference to the hub is added to these as the
hub
item before forwarding them to the constructor method
for the component. The component is then cached for subsequent use.
# calling the construct() method like this... $hub->construct( fuzzbox => { volume => 11 } ); # ...results in code equivalent to this: use Your::Module::Fuzzbox; Your::Module::Fuzzbox->new({ volume => 11, hub => $hub });
This method is installed as an auto_can handler which is called to resolved undefined methods. If the method called matches the name of a component then it calls auto_component() to generate a method to access the component. If it matches the name of a delegate method then it calls auto_delegate() to generate a delegate method.
This method generates a component method named $name
which
accesses an instance of the $module
component module.
This method generates a delegate method named $name
which
delegates to the $name
method of the $component
component.
This method returns a reference to a Badger::Config object representing the configuration for the hub. This is still marked experimental.
Andy Wardley http://wardley.org/