The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

Async - Asynchronous evaluation of Perl code (with optional timeouts)

SYNOPSIS

  my $proc = Async->new(sub { any perl code you want executed });
  
  if ($proc->ready) {
    # the code has finished executing
    if ($proc->error) {
      # something went wrong
    } else {
      $result = $proc->result;  # The return value of the code
    }
  }

  # or:
  $result = $proc->result('force completion');  # wait for it to finish
  

DESCRIPTION

Async executes some code in a separate process and retrieves the result. Since the code is running in a separate process, your main program can continue with whatever it was doing while the separate code is executing. This separate code is called an `asynchronous computation'. When your program wants to check to see if the asynchronous computation is complete, it can call the ready() method, which returns true if so, and false if it is still running.

After the asynchronous computation is complete, you should call the error() method to make sure that everything went all right. error() will return undef if the computation completed normally, and an error message otherwise.

Data returned by the computation can be retrieved with the result() method. The data must be a single string; any non-string value returned by the computation will be stringized. (See AsyncData below for how to avoid this.) If the computation has not completed yet, result() will return an undefined value.

result() takes an optional parameter, $force. If $force is true, then the calling process will wait until the asynchronous computation is complete before returning.

AsyncTimeout

  use Async;
  $proc = AsyncTimeout->new(sub {...}, $timeout, $special);

Async::Timeout implements a version of Async that has an automatic timeout. If the asynchronous computation does not complete before $timeout seconds have elapsed, it is forcibly terminated and returns a special value $special. The default special value is the string "Timed out\n".

All the other methods for AsyncTimeout are exactly the same as for Async.

AsyncData

  use Async;
  $proc = AsyncData->new(sub {...});

AsyncData is just like Async except that instead of returning a string, the asynchronous computation may return any scalar value. If the scalar value is a reference, the result() method will yield a refernce to a copy of this data structure.

The AsyncData module requires that Storable be installed. AsyncData::new will die if Storable is unavailable.

All the other methods for AsyncData are exactly the same as for Async.

WARNINGS FOR THE PROGRAMMER

The asynchronous computation takes place in a separate process, so nothing it does can affect the main program. For example, if it modifies global variables, changes the current directory, opens and closes filehandles, or calls die, the parent process will be unaware of these things. However, the asynchronous computatin does inherit the main program's file handles, so if it reads data from files that the main program had open, that data will not be availble to the main program; similarly the asynchronous computation can write data to the same file as the main program if it inherits an open filehandle for that file.

ERRORS

The errors that are reported by the error() mechanism are: those that are internal to Async itself:

        Couldn't make pipe: (reason)
        Couldn't fork: (reason)
        Read error: (reason)

If your asynchronous computation dies for any reason, that is not considered to be an `error'; that is the normal termination of the process. Any messages written to STDERR will go to the computation's STDERR, which is normally inherited from the main program, and the result() will be the empty string.

EXAMPLE

  use Async;
  sub long_running_computation {
     # This function simulates a computation that takes a long time to run
     my ($x) = @_;
     sleep 5;
     return $x+2;  # Eureka!
  }

  # Main program:
  my $proc = Async->new(sub {long_running_computation(2)}) or die;
  # The long-running computation is now executing.
  #

  while (1) {
    print "Main program:  The time is now ", scalar(localtime), "\n";
    my $e;
    if ($proc->ready) {
      if ($e = $proc->error) {
        print "Something went wrong.  The error was: $e\n";
      } else {
        print "The result of the computation is: ", $proc->result, "\n";
      }
      undef $proc;
    }
    # The result is not ready; we can go off and do something else here.
    sleep 1; # One thing we could do is to take nap.
  }

AUTHOR

Mark-Jason Dominus mjd-perl-async+@plover.com.