Upgrading to OkHttp 4
=====================

OkHttp 4.x upgrades our implementation language from Java to Kotlin and keeps everything else the
same. We’ve chosen Kotlin because it gives us powerful new capabilities while integrating closely
with Java.

We spent a lot of time and energy on retaining strict compatibility with OkHttp 3.x. We’re even
keeping the package name the same: `okhttp3`!

There are three kinds of compatibility we’re tracking:

  private static <E extends Comparable<? super E>> SortedSet<E> nullCheckedTreeSet(E[] elements) {
    SortedSet<E> set = newTreeSet();
    for (E element : elements) {
      // Explicit null check because TreeSet wrongly accepts add(null) when empty.
      set.add(checkNotNull(element));
    }
    return set;
  }

  public static class ContiguousSetGenerator extends AbstractContiguousSetGenerator {
    @Override

TFCI_BAZEL_COMMON_ARGS="--repo_env=HERMETIC_PYTHON_VERSION=$TFCI_PYTHON_VERSION --repo_env=USE_PYWRAP_RULES=True --config release_arm64_linux"
TFCI_BAZEL_TARGET_SELECTING_CONFIG_PREFIX=linux_arm64
# Note: this is not set to "--cpu", because that changes the package name
# to tensorflow_cpu. These ARM builds are supposed to have the name "tensorflow"
# despite lacking Nvidia CUDA support.
TFCI_BUILD_PIP_PACKAGE_WHEEL_NAME_ARG="--repo_env=WHEEL_NAME=tensorflow"

 * they end up depending on <i>other</i> existing poor-quality {@code hashCode} implementations,
 * including those in many JDK classes.
 *
 * <p>{@code Object.hashCode} implementations tend to be very fast, but have weak collision
 * prevention and <i>no</i> expectation of bit dispersion. This leaves them perfectly suitable for

 * they end up depending on <i>other</i> existing poor-quality {@code hashCode} implementations,
 * including those in many JDK classes.
 *
 * <p>{@code Object.hashCode} implementations tend to be very fast, but have weak collision
 * prevention and <i>no</i> expectation of bit dispersion. This leaves them perfectly suitable for

   *
   * <p>For example,
   *
   * <ul>
   *   <li>{@code [2, 4)} and {@code [5, 7)} are not connected
   *   <li>{@code [2, 4)} and {@code [3, 5)} are connected, because both enclose {@code [3, 4)}
   *   <li>{@code [2, 4)} and {@code [4, 6)} are connected, because both enclose the empty range
   *       {@code [4, 4)}
   * </ul>
   *
   * <p>Note that this range and {@code other} have a well-defined {@linkplain #span union} and

      return this.value == that.value;
    }
    return false;
  }

  @Override
  public int hashCode() {
    throw new UnsupportedOperationException();
  }

  // needed because otherwise Object.toString() calls hashCode()
  @Override
  public String toString() {
    return "DontHashMe" + value;
  }

  @Override
  public int compareTo(UnhashableObject o) {

    if (isDone()) {
      return;
    }
    try {
      result = runInterruptibly();
    } catch (Throwable t) {
      error = t;
    }
    if (error == null) {
      // The cast is safe because of the `run` and `error` checks.
      afterRanInterruptiblySuccess(uncheckedCastNullableTToT(result));
    } else {
      afterRanInterruptiblyFailure(error);
    }
  }

  abstract boolean isDone();

import org.gradle.work.DisableCachingByDefault

/**
 * A task used for cleaning up all accepted API changes. The functionality is called whenever the release process initiates "branching".
 */
@DisableCachingByDefault(because = "Not worth caching")
abstract class CleanAcceptedApiChanges extends DefaultTask {

    @PathSensitive(PathSensitivity.ABSOLUTE)
    @InputDirectory
    abstract DirectoryProperty getJsonFileDirectory()

   * detail is invisible to the caller, but subtle use of certain APIs may depend on these internal
   * structures.
   *
   * We make such subtle calls in [okhttp3.internal.ws.MessageInflater] because we try to read a
   * compressed stream that is terminated in a web socket frame even though the DEFLATE stream is
   * not terminated.
   *