NAME
   "Async::MergePoint" - resynchronise diverged control flow

SYNOPSIS
    use Async::MergePoint;

    my $merge = Async::MergePoint->new(
       needs => [ "leaves", "water" ],
    );

    my $water;
    Kettle->boil(
       on_boiled => sub { $water = shift; $merge->done( "water" ); }
    );

    my $tea_leaves;
    Cupboard->get_tea_leaves(
       on_fetched => sub { $tea_leaves = shift; $merge->done( "leaves" ); }
    );

    $merge->close(
       on_finished => sub {
          # Make tea using $water and $tea_leaves
       }
    );

DESCRIPTION
   Often in program logic, multiple different steps need to be taken that
   are independent of each other, but their total result is needed before
   the next step can be taken. In synchonous code, the usual approach is to
   do them sequentially.

   An asynchronous or event-based program could do this, but if each step
   involves some IO idle time, better overall performance can often be
   gained by running the steps in parallel. A "Async::MergePoint" object
   can then be used to wait for all of the steps to complete, before
   passing the combined result of each step on to the next stage.

   A merge point maintains a set of outstanding operations it is waiting
   on; these are arbitrary string values provided at the object's
   construction. Each time the "done()" method is called, the named item is
   marked as being complete. When all of the required items are so marked,
   the "on_finished" continuation is invoked.

   For use cases where code may be split across several different lexical
   scopes, it may not be convenient or possible to share a lexical
   variable, to pass on the result of some asynchronous operation. In these
   cases, when an item is marked as complete a value can also be provided
   which contains the results of that step. The "on_finished" callback is
   passed a hash (in list form, rather than by reference) of the collected
   item values.

   This module was originally part of the IO::Async distribution, but was
   removed under the inspiration of Pedro Melo's Async::Hooks distribution,
   because it doesn't itself contain anything IO-specific.

CONSTRUCTOR
 $merge = Async::MergePoint->new( %params )
   This function returns a new instance of a "Async::MergePoint" object.
   The %params hash takes the following keys:

   needs => ARRAY
           Optional. An array containing unique item names to wait on. The
           order of this array is not significant.

   on_finished => CODE
           Optional. CODE reference to the continuation for when the merge
           point becomes ready. If provided, will be passed to the "close"
           method.

METHODS
 $merge->close( %params )
   Allows an "on_finished" continuation to be set if one was not provided
   to the constructor.

   on_finished => CODE
           CODE reference to the continuation for when the merge point
           becomes ready.

   The "on_finished" continuation will be called when every key in the
   "needs" list has been notified by the "done()" method. It will be called
   as

    $on_finished->( %items )

   where the %items hash will contain the item names that were waited on,
   and the values passed to the "done()" method for each one. Note that
   this is passed as a list, not as a HASH reference.

   While this feature can be used to pass data from the component parts
   back up into the continuation, it may be more direct to use normal
   lexical variables instead. This method allows the continuation to be
   placed after the blocks of code that execute the component parts, so it
   reads downwards, and may make it more readable.

 $merge->needs( @keys )
   When called on an open MergePoint (i.e. one that does not yet have an
   "on_finished" continuation), this method adds extra key names to the set
   of outstanding names. The order of this list is not significant.

   This method throws an exception if the MergePoint is already closed.

 $merge->done( $item, $value )
   This method informs the merge point that the $item is now ready, and
   passes it a value to store, to be passed into the "on_finished"
   continuation. If this call gives the final remaining item being waited
   for, the "on_finished" continuation is called within it, and the method
   will not return until it has completed.

EXAMPLES
 Asynchronous Plugins
   Consider a program using "Module::Pluggable" to provide a plugin
   architecture to respond to events, where sometimes the response to an
   event may require asynchronous work. A "MergePoint" object can be used
   to coordinate the responses from the plugins to this event.

    my $merge = Async::MergePoint->new();

    foreach my $plugin ( $self->plugins ) {
       $plugin->handle_event( "event", $merge, @args );
    }

    $merge->close( on_finished => sub {
       my %results = @_;
       print "All plugins have recognised $event\n";
    } );

   Each plugin that wishes to handle the event can use its own package
   name, for example, as its unique key name for the MergePoint. A plugin
   handling the event synchonously could perform something such as:

    sub handle_event
    {
       my ( $event, $merge, @args ) = @_;
       ....
       $merge->needs( __PACKAGE__ );
       $merge->done( __PACKAGE__ => $result );
    }

   Whereas, to handle the event asynchronously the plugin can instead
   perform:

    sub handle_event
    {
       my ( $event, $merge, @args ) = @_;
       ....
       $merge->needs( __PACKAGE__ );

       sometime_later( sub {
          $merge->done( __PACKAGE__ => $result );
       } );
    }

AUTHOR
   Paul Evans <[email protected]>

You are viewing proxied material from ftp.icm.edu.pl. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.