* a {@code Map<R, Map<C, V>>}. A method call like {@code column(columnKey).get(rowKey)} still runs
 * quickly, since the row key is provided. However, {@code column(columnKey).size()} takes longer,
 * since an iteration across all row keys occurs.
 *
 * <p>Note that this implementation is not synchronized. If multiple threads access this table

 *   <li>hasNext()
 *   <li>hasNext();
 *   <li>remove();
 *   <li>next();
 * </ol>
 *
 * <p>This particular order of operations may be unrealistic, and testing all 3^5 of them may be
 * thought of as overkill; however, it's difficult to determine which proper subset of this massive
 * set would be sufficient to expose any possible bug. Brute force is simpler.
 *
 * <p>To use this class the concrete subclass must implement the {@link

 * a {@code Map<R, Map<C, V>>}. A method call like {@code column(columnKey).get(rowKey)} still runs
 * quickly, since the row key is provided. However, {@code column(columnKey).size()} takes longer,
 * since an iteration across all row keys occurs.
 *
 * <p>Note that this implementation is not synchronized. If multiple threads access this table

   * be upcast back to ImmutableMapEntry[] (or Object[], etc.), or allowed to escape the class.
   *
   * <p>The returned array has all its elements set to their initial null values. However, we don't
   * declare it as {@code @Nullable ImmutableMapEntry[]} because our checker doesn't require newly
   * created arrays to have a {@code @Nullable} element type even when they're created directly with

     * <li>If the class loader that loaded this class can be obtained, that class loader</li>
     * <li>If the system class loader can be obtained, that class loader</li>
     * </ol>
     * <p>
     * However, if both the class loader that loaded the target class and the class loader that loaded this class can be obtained,
     * and the class loader that loaded the target class is an ancestor of the class loader that loaded this class,

 - **CacheHit**
 - ConnectionReleased
 - CallEnd
 
## Cache directory

The cache directory must be exclusively owned by a single instance.

Deleting the cache when it is no longer needed can be done.  However this may delete the purpose of the cache
which is designed to persist between app restarts.

```kotlin
cache.delete()
```
 
## Pruning the Cache

   *
   * <p>Historically, {@code Supplier} instances in this library have implemented this method to
   * recognize certain cases where distinct {@code Supplier} instances would in fact behave
   * identically. However, as code migrates to {@code java.util.function}, that behavior will
   * disappear. It is best not to depend on it.
   */
  @Override
  boolean equals(@Nullable Object object);

 * information</i>.
 *
 * <h3>Invertibility</h3>
 *
 * <p>The reverse operation <b>may</b> be a strict <i>inverse</i> (meaning that {@code
 * converter.reverse().convert(converter.convert(a)).equals(a)} is always true). However, it is very
 * common (perhaps <i>more</i> common) for round-trip conversion to be <i>lossy</i>. Consider an
 * example round-trip using {@link com.google.common.primitives.Doubles#stringConverter}:
 *
 * <ol>

With rewrites, redirects, follow-ups and retries, your simple request may yield many requests and responses. OkHttp uses `Call` to model the task of satisfying your request through however many intermediate requests and responses are necessary. Typically this isn’t many! But it’s comforting to know that your code will continue to work if your URLs are redirected or if you failover to an alternate IP address.

   * @param length the length of the new array
   */
  /*
   * The new array contains nulls, even if the old array did not. If we wanted to be accurate, we
   * would declare a return type of `@Nullable T[]`. However, we've decided not to think too hard
   * about arrays for now, as they're a mess. (We previously discussed this in the review of
   * ObjectArrays, which is the main caller of this method.)
   */