///

* If the PR can be simplified in a way, you can ask for that, but there's no need to be too picky, there might be a lot of subjective points of view (and I will have my own as well 🙈), so it's better if you can focus on the fundamental things.

### Tests { #tests }

* Help me check that the PR has **tests**.

* Check that the tests **fail** before the PR. 🚨

}

var listCache sync.Map // map[string]listImports, keyed by contextName

// listSem is a semaphore restricting concurrent invocations of 'go list'. 'go
// list' has its own internal concurrency, so we use a hard-coded constant (to
// allow the I/O-intensive phases of 'go list' to overlap) instead of scaling
// all the way up to GOMAXPROCS.
var listSem = make(chan semToken, 2)

 *
 * ## OkHttpClients Should Be Shared
 *
 * OkHttp performs best when you create a single `OkHttpClient` instance and reuse it for all of
 * your HTTP calls. This is because each client holds its own connection pool and thread pools.
 * Reusing connections and threads reduces latency and saves memory. Conversely, creating a client
 * for each request wastes resources on idle pools.
 *

    size = 0;
    modCount++;
  }

  // Views

  /**
   * {@inheritDoc}
   *
   * <p>If the multimap is modified while an iteration over the list is in progress (except through
   * the iterator's own {@code add}, {@code set} or {@code remove} operations) the results of the
   * iteration are undefined.
   *
   * <p>The returned list is not serializable and does not have random access.
   */
  @Override

  When you convey a copy of a covered work, you may at your option
remove any additional permissions from that copy, or from any part of
it.  (Additional permissions may be written to require their own
removal in certain cases when you modify the work.)  You may place
additional permissions on material, added by you to a covered work,
for which you have or can give appropriate copyright permission.

   * WritableByteChannel)}:
   *
   * <ol>
   *   <li>Use sendfile(2) or equivalent. Requires that both the input channel and the output
   *       channel have their own file descriptors. Generally this only happens when both channels
   *       are files or sockets. This performs zero copies - the bytes never enter userspace.

            The type of artifact this project produces, for example {@code jar},
              {@code war},
              {@code ear},
              {@code pom}.
            Plugins can create their own packaging, and
            therefore their own packaging types,
            so this list does not contain all possible types.
          </description>
          <type>String</type>
          <defaultValue>jar</defaultValue>
        </field>

build:rbe_linux_cpu --copt=-Wno-gnu-offsetof-extensions
# Python config is the same across all containers because the binary is the same
build:rbe_linux_cpu --python_path="/usr/bin/python3"
# These you may need to change for your own GCP project.
common:rbe_linux_cpu --remote_instance_name=projects/tensorflow-testing/instances/default_instance

# Download CUDA/CUDNN redistributions to preserve the repositories cache between
# CPU and GPU builds.

     * longer be the case.)
     *
     * javax.naming.Name and java.util.concurrent.Delayed might work, but
     * they're fairly obscure, we've invented our own interface and class.
     */
    Interface a = new Impl();
    Interface b = new Impl();
    ImmutableSortedSet<Interface> set = ImmutableSortedSet.of(a, b);
    Object[] unused1 = set.toArray();

   *     Storage overview</a> to select an appropriate temporary directory (perhaps {@code
   *     context.getCacheDir()}), and create your own directory under that. (For example, you might
   *     use {@code new File(context.getCacheDir(), "directoryname").mkdir()}, or, if you need an