// reference counts drop to zero.
void ParallelTensorDeallocator(void* data) {
  delete reinterpret_cast<ParallelTensor*>(data);
}

// Used as an argument to TFE_NewCustomDeviceTensorHandle, for computing the
// number of dimensions of a parallel tensor.
int ParallelTensorNumDims(void* data, TF_Status* status) {
  const std::vector<int64_t>* shape;
  Status s = reinterpret_cast<ParallelTensor*>(data)->Shape(&shape);

    checkNonnegative(distance, "distance");
    for (long i = 0; i < distance; i++) {
      current = next(current);
      if (current == null) {
        throw new IllegalArgumentException(
            "overflowed computing offset(" + origin + ", " + distance + ")");
      }
    }
    return current;
  }

  /**
   * Returns the unique least value of type {@code C} that is greater than {@code value}, or {@code

  ASSERT_EQ(TF_OK, TF_GetCode(s)) << TF_Message(s);
  auto* arg1 = TF_AddFunctionParameter(graph_ctx, TF_FLOAT, {-1, nullptr}, s);
  ASSERT_EQ(TF_OK, TF_GetCode(s)) << TF_Message(s);

  // Create a first "Add" computing `arg0 + arg1`.
  TF_AbstractTensor* add_output1;
  {
    // Build an abstract operation, inputs and output.
    auto* add_op = TF_NewAbstractOp(graph_ctx);
    TF_AbstractOpSetOpType(add_op, "Add", s);

    checkNonnegative(distance, "distance");
    for (long i = 0; i < distance; i++) {
      current = next(current);
      if (current == null) {
        throw new IllegalArgumentException(
            "overflowed computing offset(" + origin + ", " + distance + ")");
      }
    }
    return current;
  }

  /**
   * Returns the unique least value of type {@code C} that is greater than {@code value}, or {@code

        //       -2_147_483_628 -> ...
        //   } }
        // That constant is encoded as `unaryMinus` call with the const 2147483628 of long type, while the resolved type is Int.
        // After computing the compile time constant, we need to adjust its type here.
        val expression =
            if (expectedKind != null && expectedKind != kind && value is Number) {

    ReferenceEntry<K, V> entry = getReferenceEntry(cache, key);
    if (entry != null) {
      ValueReference<K, V> valueRef = entry.getValueReference();
      // fail on strong/computing refs
      Preconditions.checkState(valueRef instanceof Reference);
      Reference<V> ref = (Reference<V>) valueRef;
      if (ref != null) {
        ref.clear();
      }
    }
  }

  /**

// NewReader then it should be set to -1.
// When size is >=0 it *must* match the amount of data provided by r.
//
// NewReader may try merge the given size, MD5 and SHA256 values
// into src - if src is a Reader - to avoid computing the same
// checksums multiple times.
// NewReader enforces S3 compatibility strictly by ensuring caller
// does not send more content than specified size.

      status_ =
          BuildImmediateExecutionContext(std::get<1>(GetParam()), &ctx_raw);
      ASSERT_EQ(errors::OK, status_.code()) << status_.message();
      immediate_execution_ctx_.reset(ctx_raw);
    }

    // Computing numerical gradients with TensorFloat-32 is numerically
    // unstable. Some forward pass tests also fail with TensorFloat-32 due to
    // low tolerances
    enable_tensor_float_32_execution(false);
  }

    ReferenceEntry<K, V> entry = getReferenceEntry(cache, key);
    if (entry != null) {
      ValueReference<K, V> valueRef = entry.getValueReference();
      // fail on strong/computing refs
      Preconditions.checkState(valueRef instanceof Reference);
      Reference<V> ref = (Reference<V>) valueRef;
      if (ref != null) {
        ref.clear();
      }
    }
  }

  /**

import org.checkerframework.checker.nullness.qual.NonNull;
import org.checkerframework.checker.nullness.qual.Nullable;

/**
 * A strategy for determining whether two instances are considered equivalent, and for computing
 * hash codes in a manner consistent with that equivalence. Two examples of equivalences are the
 * {@linkplain #identity() identity equivalence} and the {@linkplain #equals "equals" equivalence}.
 *
 * @author Bob Lee