// CHECK:   call @while_cond_10_frozen0
// CHECK:   yield
// CHECK: ^bb0([[ARGS]]):
// CHECK:   call @while_body
// CHECK:   yield
// CHECK: while_body
// CHECK: while_cond

// INLINE: tfl.while
// INLINE: ^bb0([[ARGS:.*]]):
// INLINE:   tf.Less
// INLINE:   yield
// INLINE: ^bb0([[ARGS]]):
// INLINE:   %cst_2 = arith.constant
// INLINE:   yield
// INLINE-NOT: while_body
// INLINE-NOT: while_cond

// -----

func.func @main() {
  tf_executor.graph {
    %0 = tf_executor.island {
      tf_executor.yield
    }
    tf_executor.fetch
  }
  func.return
}

// CHECK: functions must be of a single Graph with single op Islands: tf_executor.island must perfectly wrap a single op

// -----

func.func @main() {

  %1 = "tf.WhileRegion"(%arg0) ({
  ^cond(%carg0: tensor<*xf32>):
    %2 = "tf.Const"() {value = dense<true> : tensor<i1>} : () -> tensor<i1>
    "tf.Yield"(%2) : (tensor<i1>) -> ()
  }, {
  ^body(%barg0: tensor<*xf32>):
    %2 = "tf.SomeOp"(%barg0) : (tensor<*xf32>) -> tensor<*xf32>
    "tf.Yield"(%2) : (tensor<*xf32>) -> ()
  }) {is_stateless = false, shape_invariant} : (tensor<4xf32>) -> (tensor<*xf32>)

    ^bb0(%barg1: tensor<!tf_type.variant<tensor<?x1xf32>>>): // no predeceessors
      %cond = "tf.false"():()-> tensor<i1>
      "tf.Yield"(%cond) : (tensor<i1>) -> ()
  }, {
    ^bb0(%barg1: tensor<!tf_type.variant<tensor<?x1xf32>>>): // no predeceessors
    "tf.Yield"(%barg1) : (tensor<!tf_type.variant<tensor<?x1xf32>>>) -> ()

class CallableGenDependency:
    def __call__(self, value: str) -> Generator[str, None, None]:
        yield value


class AsyncCallableDependency:
    async def __call__(self, value: str) -> str:
        return value


class AsyncCallableGenDependency:
    async def __call__(self, value: str) -> AsyncGenerator[str, None]:
        yield value


class MethodsDependency:
    def synchronous(self, value: str) -> str:

      // CHECK: tf.Yield
      "tf.Yield"(%recv) : (tensor<f32>) -> ()
    }, {
      // CHECK-NOT: _xla_token_input_nodes
      %add = "tf.Add"(%arg1, %arg1) : (tensor<f32>, tensor<f32>) -> tensor<f32>

      // CHECK: tf.Yield
      "tf.Yield"(%add) : (tensor<f32>) -> ()
    }) { is_stateless = true}: (tensor<i1>) -> tensor<f32>

  func.func @main() {
    tf_executor.graph {
      %control = tf_executor.island {
        tf_executor.yield
      }
      tf_executor.fetch %control : !tf_executor.control
    }
    tf_executor.graph {
      %control = tf_executor.island {
        tf_executor.yield
      }
      tf_executor.fetch %control : !tf_executor.control
    }
    return
  }

  func.func @main() {
    tf_executor.graph {
      %control = tf_executor.island {
        tf_executor.yield
      }
      tf_executor.fetch %control : !tf_executor.control
    }
    tf_executor.graph {
      %control = tf_executor.island {
        tf_executor.yield
      }
      tf_executor.fetch %control : !tf_executor.control
    }
    return
  }

      tf_executor.yield %0 : tensor<f32>
    }
// Uses two islands for no other reason than preventing canonicalization from
// eliminating the graph entirely.
    %2:2 = tf_executor.island(%1#1) {
      %4 = "tf.opB"(%1#0) : (tensor<f32>) -> tensor<f32>
      tf_executor.yield %4 : tensor<f32>
    }
    tf_executor.fetch %2#0 : tensor<f32>
  }

        %4 = "tf.B"(%2) : (tensor<?xi32>) -> tensor<?xi32>
        tf_device.return %4 : tensor<?xi32>
      }) {device = "/device:test_device:0"} : () -> tensor<?xi32>

      // CHECK: tf_executor.yield %[[LAUNCH_OUTPUT]]
      tf_executor.yield %3 : tensor<?xi32>
    }
    tf_executor.fetch %1#0 : tensor<?xi32>
  }
  func.return %0 : tensor<?xi32>
}

// CHECK: func private @[[LAUNCH]]