Gpars (Groovy Parallel Systems) is the best solution for your concurrency problems with mutable objects. Gpars gives you a number of high-level abstractions for writing concurrent and parallel code in Groovy (map/reduce, fork/join, asynchronous closures, actors, agents, dataflow concurrency and other concepts), which can make your Java and Groovy code concurrent and/or parallel with little effort. Gpars is bundled in Groovy starting from version of 2.1.

Recently I had to write a script in Groovy, that works with collection in multi-threaded mode. The best and the easiest way for safely and correctly manage mutable state within the Application is Agent from Gpars framework. Agent completely encapsulates its internal state, the only way to pass messages to it is to use method send(). You can also use Groovy “<<” (leftShift) operator, or the implicit call() method which use overridden to use send() method. To get stored value from agent, you can use val property. When you call val property, agent waits while all scheduled tasks are finished and only then the result will be returned.

Agent agent = new Agent(new HashMap())
assert agent.val.size() == 0
agent.send {it.put("key","value")} // identical to: agent << { it.put("key","value")}
assert agent.val.size() == 1 // agent.val is executed only when all scheduled tasks are executed.
Code language: JavaScript (javascript)

Interesting feature is that you can send only code blocks (closures) but not the final result. Agent collects all the jobs in the queue and executes them in order of receipt. For better understanding may be considered such comparison: the agent – is a workshop with a set of materials, the entrance to which is only by the turn, so if you want to do something with this material you have to send the worker, to take the place in the queue.

You can read more about Gpars in the official documentation