response: Response?,
        ) {
          t.printStackTrace(System.out)
          latch.countDown()
        }
      },
    )

    check(latch.await(30, TimeUnit.SECONDS)) { "Timed out waiting for test $number to finish." }
    val endNanos = System.nanoTime()
    val tookMs = TimeUnit.NANOSECONDS.toMillis(endNanos - startNanos.get())
    println("Took ${tookMs}ms")
  }

    peer.acceptFrame() // ACK SETTINGS
    peer.acceptFrame() // PING
    peer.sendFrame().ping(true, Http2Connection.AWAIT_PING, 0)
    peer.acceptFrame() // SYN_STREAM
    peer.sendFrame().headers(false, 3, headerEntries("a", "android"))
    peer.acceptFrame() // PING
    peer.sendFrame().ping(true, Http2Connection.AWAIT_PING, 0)
    peer.acceptFrame() // DATA
    peer.acceptFrame() // RST_STREAM
    peer.play()

 * those implied by the returned execution delay.
 *
 * Cancellation
 * ------------
 *
 * Tasks may be canceled while they are waiting to be executed, or while they are executing.
 *
 * Canceling a task that is waiting to execute prevents that upcoming execution. Canceling a task
 * that is currently executing does not impact the ongoing run, but it does prevent a recurrence
 * from being scheduled.
 *

      }
      val nowMillis = TimeUnit.NANOSECONDS.toMillis(System.nanoTime())
      if (nowMillis >= timeoutMillis) break
      (this as Object).wait(timeoutMillis - nowMillis)
    }

    throw AssertionError("Timed out waiting for response to $url")
  }

  companion object {
    val TIMEOUT_MILLIS = TimeUnit.SECONDS.toMillis(10)
  }

Blocking APIs may be inefficient because you hold a thread idle while waiting on the network. Threads are expensive because they have both a memory overhead and a context-switching overhead.

#### Framed protocols

 * ordinary tests.
 *
 * <p>Failure of the expected event to occur within an implementation-defined "reasonable" time
 * period or an interrupt while waiting for the expected event will result in a {@link
 * RuntimeException}.
 *
 * <p>Here's an example that tests a {@code finalize} method:
 *
 * {@snippet :
 * final CountDownLatch latch = new CountDownLatch(1);

    taskRunner.withLock {
      assertThat(activeThreads).isEqualTo(0)
    }
  }

  /** Unblock a waiting task thread. Used by the test thread only. */
  fun interruptCoordinatorThread() {
    taskRunner.assertLockNotHeld()
    require(currentTask == TestThreadSerialTask)

    // Queue a task to interrupt the waiting coordinator.
    serialTaskQueue +=
      object : SerialTask {
        override fun start() {

 *   <li>After processing a URL ({@link #delayAfterProcessing()})</li>
 *   <li>When there are no URLs in the queue ({@link #delayAtNoUrlInQueue()})</li>
 *   <li>While waiting for new URLs to be added to the queue ({@link #delayForWaitingNewUrl()})</li>
 * </ul>
 *
 * <p>
 * Subclasses are responsible for implementing the abstract methods to define the actual delay

 * ordinary tests.
 *
 * <p>Failure of the expected event to occur within an implementation-defined "reasonable" time
 * period or an interrupt while waiting for the expected event will result in a {@link
 * RuntimeException}.
 *
 * <p>Here's an example that tests a {@code finalize} method:
 *
 * {@snippet :
 * final CountDownLatch latch = new CountDownLatch(1);

    <interactiveMode>true</interactiveMode>
    ]]></configuration>
      <description>
        <![CDATA[
      If false, flags the system to skip prompting the user for any information, or holding up the build
      waiting for any input.

      NOTE: It's also possible to switch to batch (i.e. non-interactive) mode using the '-B' command-line option.
    ]]></description>
    </expression>
  </expressions>