+ Respect the original code style: by using the same [codestyle][code-style],
  patches should only highlight the actual difference, not being disturbed by any formatting issues:
  + Only use spaces for indentation.
  + Create minimal diffs - disable on save actions like reformat source code or organize imports.
    If you feel the source code should be reformatted, create a separate PR for this change.

 * even {@code java.util.HashSet}, and places considerably less load on the garbage collector. Like
 * {@code java.util.LinkedHashSet}, it offers insertion-order iteration, with identical behavior.
 *
 * <p>This class should not be assumed to be universally superior to {@code
 * java.util.LinkedHashSet}. Generally speaking, this class reduces object allocation and memory

  public static final MediaType POSTSCRIPT = createConstant(APPLICATION_TYPE, "postscript");

  /**
   * <a href="http://tools.ietf.org/html/draft-rfernando-protocol-buffers-00">Protocol buffers</a>
   *
   * @since 15.0
   */
  public static final MediaType PROTOBUF = createConstant(APPLICATION_TYPE, "protobuf");

  /**

   * should speed up for a while, to take advantage of these resources. This is important when the
   * rate is applied to networking (limiting bandwidth), where past underutilization typically
   * translates to "almost empty buffers", which can be filled immediately.
   *
   * On the other hand, past underutilization could mean that "the server responsible for handling
   * the request has become less ready for future requests", i.e. its caches become stale, and

    otherGraph.putEdgeValue(1, 2, "valueA");
    assertThat(graph).isEqualTo(otherGraph);

    otherGraph.putEdgeValue(1, 2, "valueB");
    assertThat(graph).isNotEqualTo(otherGraph); // values differ
  }

  @Test
  public void incidentEdges_stableIncidentEdgeOrder_preservesIncidentEdgesOrder_directed() {
    graph = ValueGraphBuilder.directed().incidentEdgeOrder(ElementOrder.stable()).build();

  // except that when sortedDelegate's comparator is null, it points to a
  // non-null instance of Ordering.natural().
  // (cpovirk: Is sortedDelegate's comparator really ever null?)
  // The comparator will likely also differ because of our nullAccepting hack.
  // See the bottom of the file for more information about it.
  private final transient Comparator<? super K> comparator;

For example, a Contributor might include the Program in a commercial
product offering, Product X. That Contributor is then a Commercial
Contributor. If that Commercial Contributor then makes performance
claims, or offers warranties related to Product X, those performance
claims and warranties are such Commercial Contributor's responsibility
alone. Under this section, the Commercial Contributor would have to

	return &SkipReader{r, n}
}

// writerOnly hides an io.Writer value's optional ReadFrom method
// from io.Copy.
type writerOnly struct {
	io.Writer
}

// Copy is exactly like io.Copy but with reusable buffers.
func Copy(dst io.Writer, src io.Reader) (written int64, err error) {
	bufp := ODirectPoolMedium.Get().(*[]byte)
	defer ODirectPoolMedium.Put(bufp)
	buf := *bufp

	return io.CopyBuffer(writerOnly{dst}, src, buf)

    memory usage.
*   Fixed bfloat16 integration of LIBXSMM sparse mat-mul.
*   Improved performance and reduce memory usage by allowing ops to forward
    input buffers to output buffers and perform computations in-place.
*   Improved the performance of CPU assignment for strings.
*   Speed up matrix * vector multiplication and matrix * matrix with unknown
    shapes.

   */
  abstract long fn(long currentValue, long newValue);

  /**
   * Handles cases of updates involving initialization, resizing, creating new Cells, and/or
   * contention. See above for explanation. This method suffers the usual non-modularity problems of
   * optimistic retry code, relying on rechecked sets of reads.
   *
   * @param x the value
   * @param hc the hash code holder