Experimenting with concurrent execution I was wondering how to actually test it. The execution flow is of a side-effect nature and futures are created to wrap independent executions/processing.
Been searching for some good examples on how to properly unit test the following scenarios (foo
and bar
are the methods I wish to test):
scenario #1
def foo : Unit = { Future { doSomething } Future { doSomethingElse } } private def doSomething : Unit = serviceCall1 private def doSomethingElse : Unit = serviceCall2
Scenario motivation
foo
immediately returns but invokes 2 futures which perform separate tasks (e.g. save analytics and store record to DB). These service calls can be mocked, but what I'm trying to test is that both these services are called once I wrap them in Future
s
scenario #2
def bar : Unit = { val futureX = doAsyncX val futureY = doAsyncY for { x <- futureX y <- futureY } yield { noOp(x, y) } }
Scenario motivation
Start with long running computations that can be executed concurrently (e.g. get the number of total visitors and get the frequently used User-Agent
header to our web site). Combine the result in some other operation (which in this case Unit
method that simply throws the values)
Note I'm familiar with actors and testing actors, but given the above code I wonder what should be the most suitable approach (refactoring included)
EDIT What I'm doing at the moment
implicit value context = ExecutionContext.fromExecutor(testExecutor) def testExecutor = { new Executor { def execute(runnable : Runnable) = runnable.run } }
This ExecutionContext
implementation will not run the Future
as a separate thread and the entire execution will be done in sequence. This kinda feels like a hack but based on Electric Monk answer, it seems like the other solution is more of the same.
noOp
can be a mocked services that I wish to test if it was invoked as expected – Bivas