* request is transmitted. This is only interesting for duplex calls where the request and
   * response may be interleaved.
   *
   * Read this value only once for each enter/exit pair because its value can change.
   */
  private fun doReadTimeout() = !connection.client || sink.closed || sink.finished

  /**
   * A source that reads the incoming data frames of a stream. Although this class uses

   *
   * 2. `other[size] = null` is unsound. We could "fix" this by requiring callers to pass in an
   * array with a nullable element type. But probably they usually want an array with a non-nullable
   * type. That said, we could *accept* a `@Nullable T[]` (which, given that we treat arrays as
   * covariant, would still permit a plain `T[]`) and return a plain `T[]`. But of course that would

   *
   * <p>In addition to operating on {@code Map} instances, {@code MapJoiner} can operate on {@code
   * Multimap} entries in two distinct modes:
   *
   * <ul>
   *   <li>To output a separate entry for each key-value pair, pass {@code multimap.entries()} to a
   *       {@code MapJoiner} method that accepts entries as input, and receive output of the form
   *       {@code key1=A&key1=B&key2=C}.

corresponding type parameters have corresponding constraints.
</p>

<pre>
type Pair[A, B any] struct {
	a A
	b B
}

func (p Pair[A, B]) Swap() Pair[B, A]  { … }  // receiver declares A, B
func (p Pair[First, _]) First() First  { … }  // receiver declares First, corresponds to A in Pair
</pre>

<p>

    and
    [GPU developer Dockerfile](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tools/dockerfiles/dockerfiles/devel-gpu.Dockerfile)
    for the required packages. Alternatively, use the said
    [tensorflow/build Docker images](https://hub.docker.com/r/tensorflow/build)
    (`tensorflow/tensorflow:devel` and `tensorflow/tensorflow:devel-gpu` are no
    longer supported for development). Use TF SIG Build Dockerfiles in

   * NullPointerException} on null values returned from {@code valueFunction}, and treats nulls
   * returned from {@code mergeFunction} as removals of that row/column pair.
   *
   * @since 33.2.0 (available since 21.0 in guava-jre)
   */
  @IgnoreJRERequirement // Users will use this only if they're already using streams.
  public static <
          T extends @Nullable Object,

   * bulkGet(keys)} with a relative large number of keys can cause an excessive number of shared
   * stripes (much like the birthday paradox, where much fewer than anticipated birthdays are needed
   * for a pair of them to match). Please consider carefully the implications of the number of
   * stripes, the intended concurrency level, and the typical number of keys used in a {@code

pkg syscall (windows-386), type CertRevocationInfo struct, OidSpecificInfo uintptr
pkg syscall (windows-386), type CertSimpleChain struct, TrustListInfo uintptr
pkg syscall (windows-386), type RawSockaddrAny struct, Pad [96]int8
pkg syscall (windows-amd64), const TOKEN_ALL_ACCESS = 983295
pkg syscall (windows-amd64), type AddrinfoW struct, Addr uintptr
pkg syscall (windows-amd64), type CertChainPolicyPara struct, ExtraPolicyPara uintptr

     * call, so we can't assume that writes to the Entry have been safely published by some other
     * synchronization actions.)
     *
     * All that said: I haven't actually managed to make this particular test produce a TSAN error
     * for the field accesses in MapIteratorCache. This test *has* found other TSAN errors,

   * for sorting user-visible strings as the ordering may not match the conventions of the user's
   * locale. That is, it compares, using {@link #compare(char, char)}), the first pair of values
   * that follow any common prefix, or when one array is a prefix of the other, treats the shorter
   * array as the lesser. For example, {@code [] < ['a'] < ['a', 'b'] < ['b']}.
   *