}
    }

    return mlir::UnrankedTensorType::get(element_type);
  };

  // Below we only try and do some shape inference for "source" ops which have
  // no inputs.
  if (node.num_inputs() > 0) return default_type();

  // Do some simply inference here to get the function arguments correct for
  // this common case.
  // TODO(jpienaar): Reconsider post refactoring shape functions.

	// do not block indefinitely (e.g. apply a timeout).
	Close() error

	// NumInput returns the number of placeholder parameters.
	//
	// If NumInput returns >= 0, the sql package will sanity check
	// argument counts from callers and return errors to the caller
	// before the statement's Exec or Query methods are called.
	//
	// NumInput may also return -1, if the driver doesn't know

      IsSingleCallRegion(while_region.getBody()), while_arg_matcher);

  // All existing inputs to while region are inputs to the functional while.
  auto new_inputs = llvm::to_vector<4>(while_region.getOperands());

  // All existing results will also be generated by the functional while.
  auto new_result_types = llvm::to_vector<4>(while_region.getResultTypes());

					return err
				}
				covervarsoutfile = outputfiles[0]
				outputfiles = outputfiles[1:]
				numInputs := len(flag.Args())
				numOutputs := len(outputfiles)
				if numOutputs != numInputs {
					return fmt.Errorf("number of output files (%d) not equal to number of input files (%d)", numOutputs, numInputs)
				}
				if err := readPackageConfig(*pkgcfg); err != nil {
					return err
				}
				return nil

/**
 * Tests for SimpleGenericBloomFilter and derived BloomFilter views.
 *
 * @author Dimitris Andreou
 */
public class BloomFilterTest extends TestCase {
  private static final int NUM_PUTS = 100_000;
  private static final ThreadLocal<Random> random =
      new ThreadLocal<Random>() {
        @Override
        protected Random initialValue() {
          return new Random();
        }
      };

  const auto &inputs_ty = op.getInputs().getType();
  int n_inputs = inputs_ty.size();
  if (n_inputs < 1) return op.emitOpError() << "No inputs";

  const auto &init_values_ty = op.getInitValues().getType();
  int n_init_values = init_values_ty.size();
  if (n_init_values != n_inputs) {
    return op.emitOpError() << "Number of inputs (" << n_inputs
                            << ") is different than number of init_values ("

def LowerSoftmaxCrossEntropyWithLogitsOp : Pattern<
  (TF_SoftmaxCrossEntropyWithLogitsOp AnyRankedTensor:$features,
                                      AnyRankedTensor:$labels),
  [(TF_SumOp (TF_MulNoNanOp:$sum_input
                     (TF_LogSoftmaxOp $features),
                     (TF_NegOp $labels)),
             (TF_ConstOp (GetI64ScalarElementsAttr<-1>)),
             /*keep_dims=*/ConstBoolAttrFalse),

		posRow:    -1,
		rows:      setMRows,
		cols:      setColumns,
		colType:   setColType,
		errPos:    -1,
	}
	return cursor, nil
}

func (s *fakeStmt) NumInput() int {
	if s.panic == "NumInput" {
		panic(s.panic)
	}
	return s.placeholders
}

// hook to simulate broken connections
var hookCommitBadConn func() bool

func (tx *fakeTx) Commit() error {

def OptimizeToSoftmax : Pat<
  (TFL_DivOp (TFL_ExpOp:$exp $input),
             (TFL_SumOp:$sum $sum_input, (Arith_ConstantOp I32ElementsAttr: $axes),
                             ConstBoolAttrTrue), TFL_AF_None),
  (TFL_SoftmaxOp $input, ConstF32Attr<"1.0">),
  [(IsSame $exp, $sum_input),
   (AxesIsLastDimension $axes, $sum_input),
   (HasTwoUse $exp),
   (HasOneUse $sum)]>;

    int num_input = input_ty.getNumElements();
    int input_last_dim = input_ty.getShape().back();
    if (input_last_dim < 1) return std::nullopt;
    int num_result = result_ty.getNumElements();
    auto size_last = *(--size.value_end<APInt>());
    int32_t k = size_last.getSExtValue();
    if (num_input / input_last_dim * k != num_result) return std::nullopt;