* `%F0%9F%8D%A9`). This approach works for whitespace characters, control characters, non-ASCII
 * characters, and characters that already have another meaning in a particular context.
 *
 * Percent encoding is used in every URL component except for the hostname. But the set of
 * characters that need to be encoded is different for each component. For example, the path

  @ParameterizedTest
  @ArgumentsSource(ProtocolParamProvider::class)
  fun pingsTransmitted(
    protocol: Protocol,
    mockWebServer: MockWebServer,
  ) {
    setUp(protocol, mockWebServer)
    // Ping every 500 ms, starting at 500 ms.
    client =
      client.newBuilder()
        .pingInterval(Duration.ofMillis(500))
        .build()

    // Delay the response to give 1 ping enough time to be sent and replied to.

  // first request is issued will be included in any subsequent continued requests. This is sometimes
  // referred to as a consistent snapshot, and ensures that a client that is using limit to receive
  // smaller chunks of a very large result can ensure they see all possible objects. If objects are
  // updated during a chunked list the version of the object that was present at the time the first list
  // result was calculated is returned.

     *
     * This implements Happy Eyeballs ([RFC 6555][rfc_6555]), balancing connect latency vs.
     * wasted resources.
     *
     * Defaults to enabled, call with [fastFallback] = false to revert to 4.x behaviour.
     *
     * [rfc_6555]: https://datatracker.ietf.org/doc/html/rfc6555
     */
    fun fastFallback(fastFallback: Boolean) =
      apply {
        this.fastFallback = fastFallback
      }

The steps below show how to use this notification target in `namespace` format. The other format is very similar and is omitted for brevity.

### Step 1: Ensure Elasticsearch minimum requirements are met

    // incomplete.
    val creator = cache.edit("k1")!!
    creator.newSink(0).buffer().use {
      it.writeUtf8("Hello")
    }

    // Simulate a severe Filesystem failure on the first initialization.
    filesystem.setFaultyDelete(cacheDir / "k1.0.tmp", true)
    filesystem.setFaultyDelete(cacheDir, true)
    cache =

# Please note that MKL on MacOS is still not supported.
# If you would like to use a local MKL instead of downloading, please set the
# environment variable "TF_MKL_ROOT" every time before build.
build:mkl --define=build_with_mkl=true --define=enable_mkl=true
build:mkl --define=tensorflow_mkldnn_contraction_kernel=0
build:mkl --define=build_with_openmp=true
build:mkl -c opt