* + i) mod array.length}. This is equivalent to {@code Collections.rotate(Shorts.asList(array),
   * distance)}, but is considerably faster and avoids allocation and garbage collection.
   *
   * <p>The provided "distance" may be negative, which will rotate left.
   *
   * @since 32.0.0
   */
  public static void rotate(short[] array, int distance) {

   * + i) mod array.length}. This is equivalent to {@code Collections.rotate(Shorts.asList(array),
   * distance)}, but is considerably faster and avoids allocation and garbage collection.
   *
   * <p>The provided "distance" may be negative, which will rotate left.
   *
   * @since 32.0.0
   */
  public static void rotate(short[] array, int distance) {

   *
   * <p><b>Note:</b> in almost all circumstances, the returned {@link ImmutableList} retains a
   * strong reference to {@code this}, which may prevent the original list from being garbage
   * collected. If you want the original list to be eligible for garbage collection, you should
   * create and use a copy of the sub list (e.g., {@code
   * ImmutableList.copyOf(originalList.subList(...))}).
   */
  @Override

 *
 * <p>This implementation consumes significantly less memory than {@code java.util.LinkedHashSet} or
 * 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

 *   <li>Can be streamed without "breaking the chain": {@code foo.getBarInts().stream()...}.
 *   <li>Access to all collection-based utilities via {@link #asList} (though at the cost of
 *       allocating garbage).
 * </ul>
 *
 * <p>Disadvantages compared to {@code int[]}:
 *
 * <ul>
 *   <li>Memory footprint has a fixed overhead (about 24 bytes per instance).

 * than {@code size()}. Furthermore, this structure only depends on a fixed number of arrays; {@code
 * add(x)} operations <i>do not</i> create objects for the garbage collector to deal with, and for
 * every element added, the garbage collector will have to traverse {@code 1.5} references on
 * average, in the marking phase, not {@code 5.0} as in {@code java.util.HashSet}.
 *

#### Garbage Collector
* The garbage collector now supports custom APIs added via Custom Resource Definitions
  or aggregated API servers. The garbage collector controller refreshes periodically.
  Therefore, expect a latency of about 30 seconds between when an API is added and when
  the garbage collector starts to manage it.


#### Monitoring/Prometheus

 * from uniform), and amortized since some operations can trigger a hash table resize.
 *
 * <p>As compared with {@link java.util.LinkedHashMap}, this structure places significantly reduced
 * load on the garbage collector by only using a constant number of internal objects.
 *
 * <p>This class should not be assumed to be universally superior to {@code

	0x0043 00067 (x.go:3)	CALL	runtime.morestack_noctxt(SB)
	0x0048 00072 (x.go:3)	JMP	0
...
</pre>

<p>
The <code>FUNCDATA</code> and <code>PCDATA</code> directives contain information
for use by the garbage collector; they are introduced by the compiler.
</p>

<p>
To see what gets put in the binary after linking, use <code>go tool objdump</code>:
</p>

<pre>
$ go build -o x.exe x.go

   * Calls are leaked if the connection is tracking them but the application code has abandoned
   * them. Leak detection is imprecise and relies on garbage collection.
   */
  private fun pruneAndGetAllocationCount(
    connection: RealConnection,
    now: Long,
  ): Int {
    connection.lock.assertHeld()

    val references = connection.calls