NAME

SYNOPSIS

    use HTTP::Proxy;

    # use the default engine for your system
    my $proxy = HTTP::Proxy->new();

    # choose one
    my $proxy = HTTP::Proxy->new( engine => 'Old' );

DESCRIPTION

The role of an engine is to implement the main fork+serve loop with all the required bookkeeping. This is also a good way to test various implementation and/or try out new algorithms without too much difficulties.

METHODS

new()

Create a new engine. The parameter "engine" is used to decide which kind of engine will be created. Other parameters are passed to the underlying engine.

This method also implement the subclasses constructor (they obviously do not need the "engine" parameter).

CREATING YOUR OWN ENGINE

start()

This method should handle any initialisation required when the engine starts.

run()

This method is the main loop of the master process. It defines how child processes are forked, checked and killed.

The engine MUST have a run() method, and it will be called again and again until the proxy exits.

"$self->proxy->daemon" returns the listening socket that can "accept()" connections. The child must call "$self->proxy->serve_connections()" on the returned socket to handle actual TCP connections.

stop()

This optional method should handle any cleanup procedures when the engine stops (typically when the main proxy process is killed).

A subclass may also define a %defaults hash (with "our") that contains the default values for the fields used internaly.

METHODS PROVIDED TO SUBCLASSES

proxy()

Return the HTTP::Proxy object that runs the engine.

max_clients()

Get or set the maximum number of TCP clients, that is to say the maximum number of forked child process.

Some engines may understand a value of 0 as do not fork at all. This is what HTTP::Proxy::Engine::Legacy does.

make_accessors( @names )

Create accessors named after @names in the subclass package. All accessors are read/write. This is a utility method.

This is a class method.

AUTHOR

COPYRIGHT

LICENSE