* necessary. No observable state associated with this cache is modified until loading completes.
   * Unlike {@link #get}, this method does not throw a checked exception, and thus should only be
   * used in situations where checked exceptions are not thrown by the cache loader.
   *
   * <p>If another call to {@link #get} or {@link #getUnchecked} is currently loading the value for

   * returns the elements in the order they are returned by the collection's iterator. The returned
   * array is "safe" in that no references to it are maintained by the collection. The caller is
   * thus free to modify the returned array.
   *
   * <p>This method assumes that the collection size doesn't change while the method is running.
   *
   * <p>TODO(kevinb): support concurrently modified collections?
   *

-   Different Python versions
-   Linux, MacOS, and Windows machines (these pool definitions are internal)
-   x86 and arm64
-   CPU-only, or with NVIDIA CUDA support (Linux only), or with TPUs

## How to Test Your Changes to TensorFlow

You may check how your changes will affect TensorFlow by:

1. Creating a PR and observing the presubmit test results
2. Running the CI scripts locally, as explained below

      }
      /*
       * requireNonNull is safe because:
       *
       * - columnIterator started off pointing to an empty iterator, so we must have entered the
       *   `if` body above at least once. Thus, if we got this far, that `if` body initialized
       *   rowEntry at least once.
       *
       * - The only case in which rowEntry is cleared (during remove() below) happens only if the

   * of this method requires a corresponding call to {@link #maximumWeight(long)} prior to calling
   * {@link #build}. Weights are measured and recorded when entries are inserted into the cache, and
   * are thus effectively static during the lifetime of a cache entry.
   *
   * <p>When the weight of an entry is zero it will not be considered for size-based eviction
   * (though it still may be evicted by other means).
   *

      }
      /*
       * requireNonNull is safe because:
       *
       * - columnIterator started off pointing to an empty iterator, so we must have entered the
       *   `if` body above at least once. Thus, if we got this far, that `if` body initialized
       *   rowEntry at least once.
       *
       * - The only case in which rowEntry is cleared (during remove() below) happens only if the

        jcifs.SID[] arr = new jcifs.SID[] { s1, s2, s3 };

        // First call should resolve two (offset=0,length=2)
        doAnswer(inv -> {
            // Simulate resolve call by setting names/types and thus enabling cache population
            Object[] args = inv.getArguments();
            String server = (String) args[0];
            CIFSContext c = (CIFSContext) args[1];

   * Table entries are of class Cell; a variant of AtomicLong padded
   * to reduce cache contention on most processors. Padding is
   * overkill for most Atomics because they are usually irregularly
   * scattered in memory and thus don't interfere much with each
   * other. But Atomic objects residing in arrays will tend to be
   * placed adjacent to each other, and so will most often share
   * cache lines (with a huge negative performance impact) without

   * be performed atomically on the returned map. Additionally, {@code size} and {@code
   * containsValue} are implemented as bulk read operations, and thus may fail to observe concurrent
   * writes.
   *
   * @return a serializable concurrent map having the requested features
   */
  public <K, V> ConcurrentMap<K, V> makeMap() {
    if (!useCustomMap) {

   * requests which resulted in either successful or failed loading attempts, and requests which
   * waited for other threads to finish loading. It is thus the case that {@code missCount >=
   * loadSuccessCount + loadExceptionCount}. Multiple concurrent misses for the same key will result
   * in a single load operation.
   */
  public double missRate() {