* Thread.interrupt() (as discussed in InterruptibleTask). At the end of the day, it's
         * unlikely that cancel() will be slow, so we can probably get away with calling it while
         * holding a lock. Still, it would be nice to avoid somehow.
         */
        lock.lock();
        try {
          currentFuture.cancel(mayInterruptIfRunning);
        } finally {

      <dependency>
        <groupId>com.google.j2objc</groupId>
        <artifactId>j2objc-annotations</artifactId>
        <version>${j2objc.version}</version>
      </dependency>
      <!--
      We moved away from using dependencyManagement for test-only deps because of https://github.com/google/guava/issues/6654.
      We could probably have resumed it after https://github.com/google/guava/pull/6664.

      <dependency>
        <groupId>com.google.j2objc</groupId>
        <artifactId>j2objc-annotations</artifactId>
        <version>${j2objc.version}</version>
      </dependency>
      <!--
      We moved away from using dependencyManagement for test-only deps because of https://github.com/google/guava/issues/6654.
      We could probably have resumed it after https://github.com/google/guava/pull/6664.

   */
  private InternetDomainName ancestor(int levels) {
    ImmutableList<String> ancestorParts = parts.subList(levels, parts.size());

    // levels equals the number of dots that are getting clipped away, then add the length of each
    // clipped part to get the length of the leading substring that is being removed.
    int substringFrom = levels;
    for (int i = 0; i < levels; i++) {

   * mode}.
   *
   * <p>For the case of {@link RoundingMode#HALF_DOWN}, {@code HALF_UP}, and {@code HALF_EVEN},
   * infinite {@code double} values are considered infinitely far away. For example, 2^2000 is not
   * representable as a double, but {@code roundToDouble(BigInteger.valueOf(2).pow(2000), HALF_UP)}
   * will return {@code Double.MAX_VALUE}, not {@code Double.POSITIVE_INFINITY}.
   *

http://Goo%20\sgoo%7C|.com

# U+3000 is mapped to U+0020 (space) which is disallowed
http://GOO\u00a0\u3000goo.com

# Other types of space (no-break, zero-width, zero-width-no-break) are
# name-prepped away to nothing.
# U+200B, U+2060, and U+FEFF, are ignored
http://GOO\u200b\u2060\ufeffgoo.com  s:http p:/ h:googoo.com

# Ideographic full stop (full-width period for Chinese, etc.) should be
# treated as a dot.

       *   user task -- which means that we already got past this code (or else we exited early
       *   above))
       */
      // Unconditionally set; there is no risk of throwing away a queued task from another thread,
      // because in order for the current task to run on this executor the previous task must have
      // already started execution. Because each task on a TaskNonReentrantExecutor can only produce

         * Thread.interrupt() (as discussed in InterruptibleTask). At the end of the day, it's
         * unlikely that cancel() will be slow, so we can probably get away with calling it while
         * holding a lock. Still, it would be nice to avoid somehow.
         */
        lock.lock();
        try {
          currentFuture.cancel(mayInterruptIfRunning);
        } finally {

=== ":material-language-kotlin: Kotlin"
    ```kotlin
      private val executor = Executors.newScheduledThreadPool(1)

  // way in such a visibility issue would surface is most likely as a failure of cancel() to
  // propagate to the input. Cancellation propagation is fundamentally racy so this is fine.
  //
  // Future versions of the JMM may revise safe construction semantics in such a way that we can
  // safely publish these objects and we won't need this whole discussion.