int skipped = advance(iterator, position);
    if (!iterator.hasNext()) {
      throw new IndexOutOfBoundsException(
          "position ("
              + position
              + ") must be less than the number of elements that remained ("
              + skipped
              + ")");
    }
    return iterator.next();
  }

  /**

An interesting detail is that the percentage of the **CPU used** by each process can **vary** a lot over time, but the **memory (RAM)** normally stays more or less **stable**.

If you have an API that does a comparable amount of computations every time and you have a lot of clients, then the **CPU utilization** will probably *also be stable* (instead of constantly going up and down quickly).

This idea of **asynchronous** code described above is also sometimes called **"concurrency"**. It is different from **"parallelism"**.

**Concurrency** and **parallelism** both relate to "different things happening more or less at the same time".

But the details between *concurrency* and *parallelism* are quite different.

To see the difference, imagine the following story about burgers:

### Concurrent Burgers { #concurrent-burgers }

				}, w)
			}

			// fallback only when set.
			for {
				fd := fallback(werr)
				if fd == nil {
					break
				}
				// This fallback is only set when
				// askDisks is less than total
				// number of disks per set.
				werr = fd.WalkDir(ctx, WalkDirOptions{
					Limit:          opts.perDiskLimit,
					Bucket:         opts.bucket,
					BaseDir:        opts.path,

   * corrected sample standard deviation</a> of the values. If this dataset is a sample drawn from a
   * population, this is an estimator of the population standard deviation of the population which
   * is less biased than {@link #populationStandardDeviation()} (the unbiased estimator depends on
   * the distribution). The count must be greater than one.
   *

content:"\f59b"}.fa-laugh-wink:before{content:"\f59c"}.fa-layer-group:before{content:"\f5fd"}.fa-leaf:before{content:"\f06c"}.fa-leanpub:before{content:"\f212"}.fa-lemon:before{content:"\f094"}.fa-less:before{content:"\f41d"}.fa-less-than:before{content:"\f536"}.fa-less-than-equal:before{content:"\f537"}.fa-level-down-alt:before{content:"\f3be"}.fa-level-up-alt:before{content:"\f3bf"}.fa-life-ring:before{content:"\f1cd"}.fa-lightbulb:before{content:"\f0eb"}.fa-line:before{content:"\f3c0"}.fa-lin...

      for (int i = 0; i < params.size(); i++) {
        TypeToken<?> paramType = type.resolveType(typeVars[i]);
        // We require all @Generates methods to either be parameter-less or accept non-null
        // values for their generic parameter types.
        Object argValue = generate(paramType);
        if (argValue == null) {
          // When a parameter of a @Generates method cannot be created,

     * Luckily, our nullness checker is smart enough to realize that `convert` has @PolyNull-like
     * behavior, so it knows that `convert(a)` returns a non-nullable value, and we don't need to
     * perform even a cast, much less a runtime check.
     *
     * All that said, don't forget that everyone should call converter.convert() instead of
     * converter.apply(), anyway. If clients use only converter.convert(), then their nullness

        stream = getStream(streamId)

        if (stream == null) {
          // If we're shutdown, don't bother with this stream.
          if (isShutdown) return

          // If the stream ID is less than the last created ID, assume it's already closed.
          if (streamId <= lastGoodStreamId) return

          // If the stream ID is in the client's namespace, assume it's already closed.

   * Returns a string representation of this fluent iterable, with the format {@code [e1, e2, ...,
   * en]}.
   *
   * <p><b>{@code Stream} equivalent:</b> {@code stream.collect(Collectors.joining(", ", "[", "]"))}
   * or (less efficiently) {@code stream.collect(Collectors.toList()).toString()}.
   */
  @Override
  public String toString() {
    return Iterables.toString(getDelegate());
  }

  /**