Agent in

About Agent

While spawning a process is easy, managing its state and behavior can become very complicated. The Agent module is an abstraction to facilitate managing a shared state in an Elixir process.

When using Agent module functions it is customary to encapsulate the agent-related functions in a single module.

# A simple use of agent processes to represent a counter
defmodule Counter do
  def start_link(initial_value) do
    Agent.start(fn -> initial_value end)
  end

  def value(pid) do
    Agent.get(pid, & &1)
  end

  def increment(pid) do
    Agent.update(pid, &(&1 + 1))
  end
end

The choice to use the Agent module, or to use multiple processes at all, depends on the relationship to be modeled with the data. Agent processes are useful when a shared state must be used by multiple processes.

It is often the better choice to avoid using a separate process and use an in-process variable to hold the state in a function. Care should also be taken from blindly treating them as a global variable, as they may be manipulated by other processes, creating race conditions, or untraceable errors.

If choosing to use an agent process, the goal should be to hold a simple state, encapsulating the calls to the agent process within a module to organize the calls.

If an expensive job should be done in another process, a task process (using Task) process should be considered. If it is determined that a process should hold state and perform behaviours, a GenServer process (using GenServer) should be considered.