Newsletter sign-up
View all newsletters

Enterprise Java Newsletter
Stay up to date on the latest tutorials and Java community news posted on JavaWorld

Sponsored Links

Optimize with a SATA RAID Storage Solution
Range of capacities as low as $1250 per TB. Ideal if you currently rely on servers/disks/JBODs

JavaWorld.com > Java Standard Edition >

Understanding actor concurrency, Part 1: Actors in Erlang

A new way to think about structuring concurrent applications

By Alex Miller, JavaWorld.com, 02/24/09

  • Print
  • Feedback
.

Page 3 of 6

Messages are sent asynchronously and are buffered in an actor's mailbox. A mailbox is essentially a queue with multiple producers (other actors) and a single consumer. A particular actor is driven by receiving messages from the mailbox based on pattern matching.

Some key assumptions are built into this model:

  • Processes are cheap (in memory) and fast to create
  • Processes are small and thus large numbers of processes can be created
  • Processes are not bound to kernel threads and can be efficiently scheduled in user space
  • The scheduler requires the ability to pause and continue process execution
  • Messages can be retrieved by efficiently pattern-matching messages in the mailbox

Erlang

The actor model has most famously been associated with Erlang, a functional, dynamically typed language invented in 1986 at Ericsson. Erlang was designed for creating applications (such as telephone switches) that must run nonstop. That requires the code to be hot-swappable, distributed, and robust in the face of errors. Erlang was open-sourced in 1998 and has gained prominence in the last few years as concurrency has become a concern. A complete overview of Erlang is outside this article's scope, but I'll give you a quick introduction to the language's key aspects.

Erlang programs are compiled and run in a virtual machine, but you can also use an interactive shell to explore the basics. Erlang variables, which start with a capital letter or a _ character, can be assigned only once, unlike in most other languages. Listing 2 shows an example trace in the Erlang shell with a Value variable.

Listing 2. Erlang variables

1> Value = 4.
4
2> Value.
4
3> Value = 6.
** exception error: no match of right hand side value 6

In this trace you can see expressions being entered on the lines starting with 1>, 2>, and 3> and the response from the shell on the following lines. Line 1 assigns the value 4 to the Value variable. Line 2 returns the value of Value. Line 3 demonstrates that Value can't be reassigned to a new value because it is already bound to the value 4.

Actually, = is not an assignment operator at all in Erlang. It is really a pattern-matching operator. In the case of an unbound variable on the left-hand side, the pattern is unmatched and will bind a value from the right-hand side. But that's just the beginning of what = can do.

Words starting with lowercase letters are atoms -- fixed symbols that always represent a constant, number, or string of the same name, even across machines. A tuple is a fixed-size set of heterogenous values and is specified with { }. Here we match a variable with a tuple of atoms: 

Stooges = {larry, curly, moe}.

You can also put tuples on the left side and use variables and the pattern-matching = to extract values from a tuple on the right side: 

{Stooge1, Stooge2, Stooge3} = {larry, curly, moe}.

Here the variables Stooge1, Stooge2, and Stooge3 match the atoms larry, curly, and moe.

Lists contain a variable number of elements and are represented in [ ]. The | (pipe) is used to separate the head element from the rest of the list (similar to car and cdr in Lisp). In Listing 3 we build a list and then extract the list's head and tail with pattern matching.

  • Print
  • Feedback

Resources

More from JavaWorld