How can i add an index to each item in a java List to be processed in parallel

Question

I have a List of Rec's and I want to process them all in parallel, by calling some method foo().

foo() may flag some Rec's as erroneous and at the end all the error recs are reported by their index position (e.g. "rec 12 is not valid").
So I want to either pass the index position of each Rec to foo(), or set the index into each Rec before calling foo.
What I attempted to do (shown below) is first, sequentially, set the index to each Rec, and then in parallel call foo on each. Is there a better way to do this?

List<Rec> recs;
class Rec {
    private int recordIndex;
    public void setRecordIndex( int index) { recordIndex = index; }
    //more memebers and getter to the above...
}

//somewhere in the code
int index = 0;
recs.stream().forEach(rec -> rec.setRecordIndex(index++));
//the compiler complains about the index++ above, says it should be final
rec.parallelStream().forEach(rec -> foo(ProgressInfo, rec));

Is there a better way to do this overall? and if not, is there a way to fix the compilation error and still use stream? (instead of looping)

Is it not using a boolean to set the valid state and collecting at the end a better option? — Dakshinamurthy Karra
– Dakshinamurthy Karra, Commented Jul 25, 2018 at 6:29

Lino · Accepted Answer · 2018-07-25 06:17:36Z

2

This can be done with IntStream, also I would recommend using just one Stream:

IntStream.range(0, recs.size())
    .mapToObj(i -> {
       Rec rec = recs.get(i);
       rec.setRecordIndex(i);
       return rec;
    })
    .parallel()
    .forEach(rec -> foo(ProgressInfo, rec));

Altough it is discouraged to modify a state in streams so instead of setting the index inside the mapToObj it would probably better to return a new object. E.g. something like this:

.mapToObj(i -> {
   Rec copy = new Rec(recs.get(i)); // copy constructor
   copy.setRecordIndex(i);
   return copy;
})

Depending on the List implementation you're using (ArrayList performs better than LinkedList when using index-access), you could also use following snippet. But the use of peek in productive code is a bit controversial.

IntStream.range(0, recs.size())
   .peek(i -> recs.get(i).setRecordIndex(i))
   .parallel()
   .forEach(i -> foo(ProgressInfo, recs.get(i));

edited Jul 25, 2018 at 6:17

answered Jul 25, 2018 at 6:11

Lino

20k6 gold badges55 silver badges73 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Holger Over a year ago

There is no fundamental difference in the snippets regarding the List implementation, all of your variants call get(int), Whether you run it once per element or twice as in your last snippet, doesn’t make much difference. Regarding the side effects, there’s also not much difference between doing them in a map operation or within a peek. Creating a new object like in your second variant is the cleanest option, but keep in mind that the OP said, passing the index to foo is also ok, i.e. IntStream.range(0, recs.size()) .parallel() .forEach(ix -> foo(progressInfo, ix, recs.get(ix)))…

ernest_k · Accepted Answer · 2018-07-25 06:09:28Z

1

You can use an IntStream:

IntStream.range(0, recs.size())
         .forEach(i -> recs.get(i).setRecordIndex(i));

That simply uses a stream of numbers from 0 to recs.size() - 1, and calls recs.get for each of number.

answered Jul 25, 2018 at 6:09

ernest_k

45.5k5 gold badges58 silver badges107 bronze badges

Collectives™ on Stack Overflow

How can i add an index to each item in a java List to be processed in parallel

2 Answers 2

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related