import org.gradle.internal.operations.RunnableBuildOperation;

import java.util.ArrayList;
import java.util.Collection;
import java.util.List;
import java.util.function.Supplier;

/**
 * Executor for batches of intermediate build actions that can be initiated
 * from the top-level build action or from model builders.
 * <p>
 * All actions are always executed, and <b>may run in {@link #isParallel() parallel}</b>.

            thread.block()
            stream.received(bytes)
        }

        then:
        1 * dispatch.onOutput({ it instanceof ReadStdInEvent }) >> { instant.requested }
    }

    def "can read multiple batches of input from client"() {
        def dispatch = Mock(OutputEventListener)
        def stream = new StdInStream(dispatch)
        def text1 = "some text"
        def bytes1 = text1.bytes
        def text2 = "more"

// RUN: tf-tfrt-opt -split-input-file -tfrt-reconfig-batch-op="tfrt-min-num-batch-threads=2 tfrt-min-max-enqueued-batches=3" %s | FileCheck %s --dump-input=always

// -----

// The num_batch_threads is lowered bound to 2 from the original attribute of 1

// CHECK-LABEL: func private @batched_function
func.func private @batched_function(%arg0: tensor<1x3xf32>) -> tensor<1x3xf32> {
  %2 = "tf.Identity"(%arg0) : (tensor<1x3xf32>) -> tensor<1x3xf32>

  // The TFLite fused op does not support batching yet.
  // TODO(b/158709815): Add support for batches with padded NMS.
  auto boxes_type =
      mlir::dyn_cast<RankedTensorType>(func_.getFunctionType().getInput(0));
  if (boxes_type == nullptr || !boxes_type.hasRank() ||
      boxes_type.getRank() != 2) {
    return func_.emitWarning() << "TFLite does not support batched input for "
                                  "non_max_suppression_padded";
  }

    ): List<SmallSubprojectBucket> {
        // splitIntoBuckets() method expects us to split large element into N elements,
        // but we want to have a single bucket with N batches.
        // As a workaround, we repeat the bucket N times, and deduplicate the result at the end
        val resultIncludingDuplicates = splitIntoBuckets(
            LinkedList(subProjectTestClassTimes),

	lastTime traceTime // when we wrote the last event
	pos      int       // next write offset in arr
	lenPos   int       // position of batch length value
}

// traceBuf is per-M tracing buffer.
//
// TODO(mknyszek): Rename traceBuf to traceBatch, since they map 1:1 with event batches.
type traceBuf struct {
	_ sys.NotInHeap
	traceBufHeader

        failures.empty
    }

    def "executes tasks in a single batch in name order"() {
        Task a = task("a")
        Task b = task("b")
        Task c = task("c")

        when:
        populateAndExecute([b, c, a])

        then:
        executedTasks == [a, b, c]
        failures.empty
    }

    def "executes batches in order added"() {
        Task a = task("a")
        Task b = task("b")

// usage, and minimal GC pressure. It allocates events in batches so that even
// though we have to load the entire trace into memory, the conversion process
// shouldn't result in a doubling of memory usage, even if all converted events
// are kept alive, as we free batches once we're done with them.
//
// The conversion process is lossless.

package trace

import (
	"errors"

// a placeholder for a differently encoded event that's actually present in the experimental data.
type ExperimentalData struct {
	// Batches contain the actual experimental data, along with metadata about each batch.
	Batches []ExperimentalBatch
}

// ExperimentalBatch represents a packet of unparsed data along with metadata about that packet.
type ExperimentalBatch struct {

				},
				{
					delay: 100 * time.Millisecond,
				},
			},
			finalDelay:       3 * time.Second,
			wantRequestCount: 1,
		},
		{
			name:        "three adds in two batches",
			batchPeriod: 1 * time.Second,
			adds: []podAdd{
				{
					// endpoints.Run needs ~100 ms to start processing updates.
					delay: 200 * time.Millisecond,
				},
				{