// CHECK-NEXT: %[[INDEX_INIT:.*]] = mhlo.constant dense<0> : tensor<i32>
  // CHECK: %[[SHAPE:.*]] = shape.shape_of %arg0 : tensor<3x7xi64> -> tensor<2xindex>
  // CHECK: %[[INDEX:.*]] = "mhlo.dynamic_iota"(%[[SHAPE]]) <{iota_dimension = 0 : i64}> : (tensor<2xindex>) -> tensor<3x7xi32>
  // CHECK: %[[REDUCE:.*]]:2 = mhlo.reduce(%arg0 init: %[[INIT]]), (%[[INDEX]] init: %[[INDEX_INIT]])

// CHECK:         return %[[VAL_1]] : tensor<?xi32>
func.func @dynamic_iota_i32_1d(%arg0: tensor<1xi32>) -> tensor<?xi32> {
  %0 = "mhlo.dynamic_iota"(%arg0) <{iota_dimension = 0 : i64}> : (tensor<1xi32>) -> tensor<?xi32>
  func.return %0 : tensor<?xi32>
}

// CHECK-LABEL: func @dynamic_iota_f32_1d(

    return success();
  }
};

// Converts RangeOp for cases with the length is a dynamic value. The shape of
// the resulting tensor computed, then the start and delta is used with the
// dynamic_iota value to compute the final range value.
//
// For example, the resulting range op value:
//   %range = "tf.range"(%start, %limit, %delta)
//
// Is converted to the following.