How do you code doing 16 jobs 51 asyncs/launches with a for loop and the many not get messed up?

Hi @Eric Womer did you read and understand @Sam’s answer? https://kotlinlang.slack.com/archives/C1CFAFJSK/p1674648841566549?thread_ts=1674647885.585549&cid=C1CFAFJSK I’d suggest first fixing that and then trying again. If you need help in understanding or fixing the race condition, feel free to ask.

@uli Basically I need to get the out put as a var, probably a list, from the deferred, correct?

looked a bit at your code. and not sure what you are trying to do… you want to do 826 iterations and let’s say, you have 4 cores. Then you would start 826/4=206 jobs? I would expect you want to start 4 jobs with 206 iterations each.

I rewrote it that way yes, but for me in android stuidio it was 16 cores so 16 jobs 51 iterations and one last iteration of 10 (the remainder)

no. 51 jobs, 16 iterations

The 826 are ids that need to be passed to a method.

ah, corect

so the outside for loop would go 16 loops, and the inner would do 51

So the 1st issue is that your upper bound in this for loop

for (j in i..forCount) {

changes while the loop still itterates.

yup

This can be fixed like this:

val upperBound = forCount
async(Dispatchers.Default) {
  for (j in i..upperBound) {

Would that work, when forCount is still being changed?

After that you will probably run into an issue with concurrent list modification. So yes, you need to return the lists as the deferred from async and then join the partial lists

yeah, through an theList.add(defferedList.await() /* or would it be awaitAll()? *?)

This can be done very elegant with a map/fold instead of a for loop along these lines:

(1 .. (count - remainder) step cores).map { i-> 
  ...
  ...
  async { 
    // innerLoop
  }
}.awaitAll().flatten()

And it might be mathematically easier to count through the number of jobs and then calculate the indices belonging to the job:

(0 until cores).map { core->
  async {
    val i0 = quotent * core
    val i1 = quotent * (core + 1) - 1
    ...
  }
}

So manually add each individual call to the async block?