Provider API, domain object containers) when defining configurable parts of a plugin (tasks, extensions, domain objects).  The Provider API provides a consistent way to set conventions, wire related configuration together (extension <- domain object <- task) and avoid evaluation ordering problems.

The Gradle codebase has evolved over time and has a mixture of simple getter/setter methods, Provider API and things in between.  It can be hard to follow best practices because Gradle provided types...

  ASSERT_TRUE(TF_GetCode(status.get()) == TF_OK) << TF_Message(status.get());

  std::vector<std::string> devices{
      "/job:localhost/replica:0/task:0/device:CPU:0",
      "/job:localhost/replica:0/task:0/device:CPU:1"};
  ParallelDevice parallel_device(std::move(devices));
  std::unique_ptr<TFE_Op, decltype(&TFE_DeleteOp)> handle_op(

        enabled = false // don't publish normalized local version to remote repository when using 'publish' lifecycle task
    }
    tasks.named("publishGradleDistributionPublicationToLocalRepository") {
        enabled = false // this should not be used so we disable it to avoid confusion when using 'publish' lifecycle task
    }
    val localPublish = project.tasks.named("publishLocalPublicationToLocalRepository") {
        doFirst {

   * identify deadlocks.
   */
  @VisibleForTesting
  static final class Blocker extends AbstractOwnableSynchronizer implements Runnable {
    private final InterruptibleTask<?> task;

    private Blocker(InterruptibleTask<?> task) {
      this.task = task;
    }

    @Override
    public void run() {}

    private void setOwner(Thread thread) {
      super.setExclusiveOwnerThread(thread);
    }

/**
 * An abstract {@code ScheduledExecutorService} that allows subclasses to {@linkplain
 * #wrapTask(Callable) wrap} tasks before they are submitted to the underlying executor.
 *
 * <p>Note that task wrapping may occur even if the task is never executed.
 *
 * @author Luke Sandberg
 */
@J2ktIncompatible
@GwtIncompatible
@ElementTypesAreNonnullByDefault
abstract class WrappingScheduledExecutorService extends WrappingExecutorService

  }

  /**
   * Submits a blocking task for the given {@link Future} to provide {@link ListenableFuture}
   * functionality.
   *
   * <p><b>Warning:</b> If the input future does not already implement {@code ListenableFuture}, the
   * returned future will emulate {@link ListenableFuture#addListener} by submitting a task to the
   * given executor at the first call to {@code addListener}. The task must be started by the

  TFE_OpReset(reused_op.get(), "Identity",
              "/job:localhost/replica:0/task:0/device:CPU:0", status.get());
  ASSERT_TRUE(TF_GetCode(status.get()) == TF_OK) << TF_Message(status.get());
  ASSERT_EQ(tensorflow::string(TFE_OpGetDevice(reused_op.get(), status.get())),
            tensorflow::string("/job:localhost/replica:0/task:0/device:CPU:0"));
  ASSERT_TRUE(TF_GetCode(status.get()) == TF_OK) << TF_Message(status.get());
}

It's called "asynchronous" because the computer / program doesn't have to be "synchronized" with the slow task, waiting for the exact moment that the task finishes, while doing nothing, to be able to take the task result and continue the work.

import model.CIBuildModel
import model.Stage

class SmokeTests(model: CIBuildModel, stage: Stage, testJava: JvmCategory, id: String, task: String = "smokeTest", splitNumber: Int = 1) : OsAwareBaseGradleBuildType(os = Os.LINUX, stage = stage, init = {
    id("${model.projectId}_SmokeTest_$id")
    name = "Smoke Tests with 3rd Party Plugins ($task) - ${testJava.version.name.toCapitalized()} Linux"

    params {
        text(
            testTaskParameterName,
            ":core:embeddedIntegTest",
            display = ParameterDisplay.PROMPT,
            allowEmpty = false,
            description = "The test task you want to run"
        )
        text(
            testNameParameterName,
            """org.gradle.api.tasks.CachedTaskExecutionIntegrationTest.outputs*are*correctly*loaded*from*cache""",