assertOnError func(t *testing.T, e error)

		// TODO: Some failing test cases are included for completeness. The next library
		// minor version should allow them all to be fixed. In the meantime, this field
		// explains the behavior reason for a particular failure.
		fixme string
	}

	// Test cases are grouped by the kind of the CBOR data item being decoded, as enumerated in
	// https://www.rfc-editor.org/rfc/rfc8949.html#section-2.

lifetime of a claim. Each ResourceHandle contains data that should be passed to a specific kubelet plugin once it lands on a node. This data is returned by the driver after a successful allocation and is opaque to Kubernetes. Driver documentation may explain to users how to interpret this data if needed.\n\nSetting this field is optional. It has a maximum size of 32 entries. If null (or empty), it is assumed this allocation will be processed by a single kubelet plugin with no ResourceHandle data attached....

 *
 * <h3>Additional documentation</h3>
 *
 * <p>See the Guava User Guide for the {@code common.graph} package (<a
 * href="https://github.com/google/guava/wiki/GraphsExplained">"Graphs Explained"</a>) for
 * additional documentation, including:
 *
 * <ul>
 *   <li><a
 *       href="https://github.com/google/guava/wiki/GraphsExplained#equals-hashcode-and-graph-equivalence">

	msgNow
	msgFuture
)

// mergeLatestMem is like mergeLatest but is only concerned with
// updating the in-memory copy of the latest tree head (c.latest)
// not the configuration file.
// The when result explains when msg happened relative to our
// previous idea of c.latest:
// msgPast means msg was from before c.latest,
// msgNow means msg was exactly c.latest, and

   * ImmutableMultiset}. Elements are sorted by the specified comparator.
   *
   * <p><b>Warning:</b> {@code comparator} should be <i>consistent with {@code equals}</i> as
   * explained in the {@link Comparator} documentation.
   *
   * @since 21.0
   */
  public static <E> Collector<E, ?, ImmutableSortedMultiset<E>> toImmutableSortedMultiset(
      Comparator<? super E> comparator) {

				op += byteSizeSuffix(inst.DataSize / 8)
			} else if inst.Mode == 64 {
				op += "q"
			} else {
				op += byteSizeSuffix(inst.MemBytes)
			}

		default:
			if isFloat(inst.Op) {
				// I can't explain any of this, but it's what libopcodes does.
				switch inst.MemBytes {
				default:
					if (inst.Op == FLD || inst.Op == FSTP) && isMem(inst.Args[0]) {
						op += "t"
					}
				case 4:

	// plugin once it lands on a node. This data is returned by the driver
	// after a successful allocation and is opaque to Kubernetes. Driver
	// documentation may explain to users how to interpret this data if needed.
	//
	// Setting this field is optional. It has a maximum size of 32 entries.
	// If null (or empty), it is assumed this allocation will be processed by a

// Inside a single-quoted block of text, a repeated single quote indicates
// a literal single quote, as in:
//
//	'Don''t communicate by sharing memory.'
//
// A line beginning with # is a comment and conventionally explains what is
// being done or tested at the start of a new section of the script.
//
// Commands are executed one at a time, and errors are checked for each command;

Any variations across machines will result in cache misses.

Never re-release a non-changing binary dependency with the same version number but different contents: if this happens with a plugin dependency, you will never be able to explain why you don’t see cache reuse between machines (it’s because they have different versions of that artifact).

  // plugin once it lands on a node. This data is returned by the driver
  // after a successful allocation and is opaque to Kubernetes. Driver
  // documentation may explain to users how to interpret this data if needed.
  //
  // Setting this field is optional. It has a maximum size of 32 entries.
  // If null (or empty), it is assumed this allocation will be processed by a