Question

I have a question to Groovy threads.

My task is to translate each file in a given directory in a certain manner
and place the resulting output in a file in an other directory.

I wrote the following code, which works:

static def translateDir(fromDir, targetDir) {
    def allFiles = new File(fromDir).listFiles()
    def numFiles = allFiles.length

    for (i in 0..(numFiles - 1))
        translate(allFiles[i].getAbsolutePath(), targetDir)
}

Now, I tried to parallelize this code like this:

static def translateDir(fromDir, targetDir) {
    def allFiles = new File(fromDir).listFiles()
    def numFiles = allFiles.length
    def numCores = Runtime.getRuntime().availableProcessors()

    for (i in 0..(numCores - 1)) {
        println("Thread " + i + "starting")
        Thread.start {
            for (def j = i; j < numFiles; j += numCores) {
                println("j = " + j) 
                translate(allFiles[j].getAbsolutePath(), targetDir)
            }
        }
    }
}

which does not work and provides the output:

Thread 0 starting
Thread 1 starting
Thread 2 starting
Thread 3 starting

nunCores is 4, and numFiles is 3 in my test case.
What is going on here?

Answer 1

ok, 2 things:

1. Inside of the thread’s implicit run() method, you are referencing the ‘i’ variable. I’d have to step it through a debugger to see exactly what happens, but technically, you shouldn’t even have access to ‘i’ there, because it’s not final. So, I’d suggest creating a Runnable object to which you pass ‘i’ in the constructor. Then start the runnable on the thread.

2. Groovy has great concurrency support – no reason to roll your own like this. Take a look at GPars