Question

我想要并行修改或克隆/转换预先排序的树项目列表中的项目。也就是列表遍历从列表中的随机项开始，表示子树遍历（深度，父指针，descendantSize和angleStart / angleEnd可用于径向视图）。

我的代码目前看起来像：

items = ((AbsModel<SunburstContainer, SunburstItem>)model).listIterator(pHitItemIndex);
items.next();
for (int i = pHitItemIndex + 1; i < endIndex && items.hasNext(); i++) {
  final SunburstItem oldChild = items.next();
  final SunburstItem newChild = new SunburstItem(oldChild);
  final float newStartAngle =
    (oldChild.getAngleStart() - oldRoot.getAngleStart()) * angleFactor;
  LOGWRAPPER.debug("angleStart: " + newStartAngle);
  newChild.setAngleStart(newStartAngle);
  float newEndAngle =
    (oldChild.getAngleEnd() - oldChild.getAngleStart()) * angleFactor
     + newChild.getAngleStart();
  LOGWRAPPER.debug("angleEnd: " + newEndAngle);
  if (newEndAngle > PConstants.TWO_PI) {
    newEndAngle = PConstants.TWO_PI;
  }
  newChild.setAngleEnd(newEndAngle);
  newChild.setAngleCenter(newChild.getAngleStart()
    + ((newChild.getAngleEnd() - newChild.getAngleStart()) / 2f));
  newChild.setIndexToParent(oldChild.getIndexToParent() - pHitItemIndex);
  int newDepth = 0;
  LOGWRAPPER.debug("child depth: " + oldChild.getDepth());
  LOGWRAPPER
    .debug("parent depth: " + model.getItem(oldChild.getIndexToParent()).getDepth());
  final int parentDepth = model.getItem(oldChild.getIndexToParent()).getDepth();
  if ((parentDepth + 1) != oldChild.getDepth()) {
    newDepth = oldDepthMax + 2;
  } else {
    newDepth = newItems.get(newChild.getIndexToParent()).getDepth() + 1;
  }
  if (newDepth > depthMax) {
    depthMax = newDepth;
  }
  LOGWRAPPER.debug("newDepth: " + newDepth);
  newChild.setDepth(newDepth);
  newChild.update(mMappingMode, mBuffer);
  newItems.add(newChild);
}

我想简单地将（子）列表（即[pHitItemIndex + 1，endIndex]）划分为不同的部分，并将Callables提交给ExecutorService。目前整个算法是在提交给ExecutorService的Callable中执行的，但我想我也可以简单地将partition / Callables提交给同一个ExecutorService。我想知道我是否会获得使用actor的优势，例如使用Akka或使用ForkJoin框架。

Answer 1

听起来像Scala中的并行集合可以提供帮助的问题。

并行化预订树

1 个答案: