} catch (e: ClassNotFoundException) {
      null
    } catch (e: RuntimeException) {
      // Throws InaccessibleObjectException (added in JDK9) on JDK 17 due to
      // JEP 403 Strongly Encapsulate JDK Internals.
      if (e.javaClass.name != "java.lang.reflect.InaccessibleObjectException") {
        throw e
      }

      null
    }
  }

  /**

    fun systemPropertiesCleared() = mode.toTrackingMode().systemPropertiesCleared()

    private
    inner class ModeHolder {
        // ModeHolder encapsulates concurrent mode updates.
        private
        var mode: TrackerMode = Initial()

        private
        inline fun <T : TrackerMode> setMode(transition: (TrackerMode) -> T): T {
            synchronized(this) {

  // Guarantee all functions have one use, which enables more exact shape
  // inference.
  pm.addPass(mlir::TF::CreateGuaranteeAllFuncsOneUsePass());
  pm.addPass(mlir::TF::CreateTFShapeInferencePass());
  // Encapsulate PartitionedCall ops within a cluster so that the composite
  // resource ops can be decomposed.
  pm.addPass(tensorflow::tf2xla::internal::CreateXlaClusterFormationPass());

    graph_->set_versions(graph_in->versions());
  }

  // TODO(b/116981129): Enhance how the device for the encapsulated subgraph is
  // determined. In case of hard placement, ensure all the encapsulated nodes
  // have the same requested device, which in turn will be the requested device
  // for the entire encapsulated subgraph. In case of soft placement, use a
  // deterministic approach to fill in the requested device. Handle co-location

import org.gradle.internal.model.ModelContainer;

import javax.annotation.Nullable;
import java.util.Collection;
import java.util.HashSet;
import java.util.Set;

/**
 * Encapsulates all logic required to build a {@link LocalConfigurationMetadata} from a
 * {@link ConfigurationInternal}. Utilizes caching to prevent unnecessary duplicate conversions
 * between DSL and internal metadata types.
 */

----
=====

[[sec:convention_plugins]]
== Share logic using convention plugins

Gradle's recommended way of organizing build logic is to use its plugin system.

We can write a plugin that encapsulates the build logic common to several subprojects in a project.
This kind of plugin is called a *convention plugin*.

	// maps function to index in slice above (index acts as function ID)
	funcTable map[fnfile]uint32

	// A table storing coverage counts for each coverable unit.
	unitTable map[extcu]uint32
}

// extcu encapsulates a coverable unit within some function.
type extcu struct {
	fnfid uint32 // index into p.funcs slice
	coverage.CoverableUnit
}

// fnfile is a function-name/file-name tuple.
type fnfile struct {
	file  string

      TF_GUARDED_BY(global_mu_);
};

// Builds OpKernel registrations on 'device' for the JIT operators
// registered on 'jit_device'. Returns ownership of a XlaDeviceOpRegistrations
// object that encapsulates the kernel registrations.
struct XlaDeviceOpRegistrations {
  std::vector<std::unique_ptr<kernel_factory::OpKernelRegistrar>>
      op_kernel_registrars;
};

XlaDeviceOpRegistrations* RegisterXlaDeviceKernels(

import java.util.HashMap;
import java.util.LinkedHashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;

import static com.google.common.base.Strings.nullToEmpty;

/**
 * Encapsulates all logic required to extract data from a {@link SoftwareComponentInternal} in order to
 * transform it to a representation compatible with Ivy.
 */
public class IvyComponentParser {

    @VisibleForTesting

// notification with the file descriptor. Failing to do this would
// mean each notification will prevent concurrent code using the
// same file descriptor in parallel.
//
// The logic dealing with re-associations is encapsulated in
// runtime·netpollupdate. This function takes care to associate the
// descriptor only with the subset of events that were previously
// part of the association, except the one that just happened. We