map.put("bar", 4);
    assertEquals("[bar=1, foo=2, bar=3, bar=4]", map.entries().toString());
    assertThat(map.keys()).containsExactly("bar", "foo", "bar", "bar").inOrder();
    map.keys().remove("bar"); // bar is no longer the first key!
    assertEquals("{foo=[2], bar=[3, 4]}", map.toString());
  }

  public void testLinkedValues() {
    Multimap<String, Integer> map = create();
    map.put("bar", 1);
    map.put("foo", 2);

  // whatever warnings might come along with it. :(

  /**
   * Returns an immutable list containing the given elements, in order.
   *
   * <p>The array {@code others} must not be longer than {@code Integer.MAX_VALUE - 12}.
   *
   * @throws NullPointerException if any element is null
   * @since 3.0 (source-compatible since 2.0)
   */

 *       ImmutableList#subList} may retain a reference to the entire data set, preventing it from
 *       being garbage collected. If some of the data is no longer reachable through other means,
 *       this constitutes a memory leak. Pass the view collection to the appropriate {@code copyOf}
 *       method to obtain a correctly-sized copy.

 * done by determining whether the domain ended with a {@linkplain #isPublicSuffix() public suffix}
 * but was not itself a public suffix. However, this test is no longer accurate. There are many
 * domains which are both public suffixes and addressable as hosts; {@code "uk.com"} is one example.
 * Using the subset of public suffixes that are {@linkplain #isRegistrySuffix() registry suffixes},

   * Comparisons between the two should take this into account.
   *
   * <p>Fingerprint2011() is a form of Murmur2 on strings up to 32 bytes and a form of CityHash for
   * longer strings. It could have been one or the other throughout. The main advantage of the
   * combination is that CityHash has a bunch of special cases for short strings that don't need to

      } catch (_: FileNotFoundException) {
        // A file must have been deleted manually!
        for (source in sources) {
          source.closeQuietly()
        }
        // Since the entry is no longer valid, remove it so the metadata is accurate (i.e. the cache
        // size.)
        try {
          removeEntry(this)
        } catch (_: IOException) {
        }
        return null
      }

        indexUpdater.start();

        // Let it run briefly
        ThreadUtil.sleep(100);

        // Stop the indexer
        indexUpdater.setFinishCrawling(true);

        // Wait for completion with longer timeout
        indexUpdater.join(3000);

        assertFalse(errorOccurred.get());
        // Thread may still be alive in test environment, just check no errors

        .addEqualityGroup(Range.all(), Range.all())
        .addEqualityGroup("Phil")
        .testEquals();
  }

  @GwtIncompatible // TODO(b/148207871): Restore once Eclipse compiler no longer flakes for this.
  @J2ktIncompatible // TODO(b/148207871): Likewise, or once J2KT uses javac instead of Eclipse.
  public void testLegacyComparable() {

 * 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

 * verbosity, and more, from the {@code get} and {@code set} methods. {@code Monitor} implements the
 * same efficient signaling as we had to hand-code in the {@code ReentrantLock} version above.
 * Finally, the programmer no longer has to hand-code the wait loop, and therefore doesn't have to
 * remember to use {@code while} instead of {@code if}.
 *
 * {@snippet :
 * public class SafeBox<V> {
 *   private V value;