* exactly what we've benchmarked, and we certainly haven't benchmarked as far back as JDK 8.
       * (We also haven't benchmarked under Android. We continue to use UnsafeAtomicHelper there so
       * that we don't change the performance there, for better or for worse.) Fortunately, JVM
       * users will typically use guava-jre, not guava-android, and guava-jre uses the VarHandle
       * implementation when possible.
       */
      try {

					var changed bool
					clearField := true
					for i, sz := range version.ObjectV2.PartActualSizes {
						if len(version.ObjectV2.PartIndices) > i {
							// 8<<20 is current 'compMinIndexSize', but we detach it in case it should change in the future.
							if sz <= 8<<20 && len(version.ObjectV2.PartIndices[i]) > 0 {
								changed = true
								version.ObjectV2.PartIndices[i] = nil

    /** SMB2 echo/keepalive command */
    protected static final short SMB2_ECHO = 0x000D;
    /** SMB2 query directory command */
    protected static final short SMB2_QUERY_DIRECTORY = 0x000E;
    /** SMB2 change notify command */
    protected static final short SMB2_CHANGE_NOTIFY = 0x000F;
    /** SMB2 query info command */
    protected static final short SMB2_QUERY_INFO = 0x0010;
    /** SMB2 set info command */

These last **three consecutive releases** are independent so that you can **migrate gradually**:

* First to FastAPI `0.68.2`, with no breaking changes, but upgrading all the sub-dependencies.
* Next to FastAPI `0.69.0`, which upgrades Starlette to `0.15.0`, with AnyIO support, and a higher chance of having breaking changes in your code.

     * for the field accesses in MapIteratorCache. This test *has* found other TSAN errors,
     * including in MapRetrievalCache, so I'm not sure why this one is different. I did at least
     * confirm that my change to MapIteratorCache fixes the TSAN error in the (larger) test it was
     * originally reported in.
     */
    for (Future<?> future : futures.build()) {
      future.get();
    }
    executor.shutdown();
  }

   * happens subsequently. As long as the corresponding key-value mapping is not removed from the
   * multimap, {@link Entry#getValue} returns the value from the multimap, which may change over
   * time, and {@link Entry#setValue} modifies that value. Removing the mapping from the multimap
   * does not alter the value returned by {@code getValue()}, though a subsequent {@code setValue()}

import java.util.function.Function;
import java.util.stream.Collector;
import java.util.stream.Collectors;
import org.jspecify.annotations.Nullable;

/**
 * A {@link Map} whose contents will never change, with many other important properties detailed at
 * {@link ImmutableCollection}.
 *
 * <p>See the Guava User Guide article on <a href=

import java.util.function.Function;
import java.util.stream.Collector;
import java.util.stream.Collectors;
import org.jspecify.annotations.Nullable;

/**
 * A {@link Map} whose contents will never change, with many other important properties detailed at
 * {@link ImmutableCollection}.
 *
 * <p>See the Guava User Guide article on <a href=

    }
    if (range.hasLowerBound() && range.hasUpperBound()) {
      return set.subSet(
          range.lowerEndpoint(),
          range.lowerBoundType() == BoundType.CLOSED,
          range.upperEndpoint(),
          range.upperBoundType() == BoundType.CLOSED);
    } else if (range.hasLowerBound()) {
      return set.tailSet(range.lowerEndpoint(), range.lowerBoundType() == BoundType.CLOSED);
    } else if (range.hasUpperBound()) {

    threadPool.awaitTermination(300, SECONDS);

    // Since we're not doing any more cache operations, and the cache only expires/evicts when doing
    // other operations, the cache and the removal queue won't change from this point on.

    // Verify that each received removal notification was valid
    for (RemovalNotification<String, String> notification : removalListener) {