original.withRole("role", false).getOutputDirectory(new File(".")) == new File(".", "parent/role/dim1")
        original.withRole("role", true).getOutputDirectory(new File(".")) == new File(".", "parent/dim1")
        original.withBinaryName("binaryName").getOutputDirectory(new File(".")) == new File(".", "parent/binaryName")
    }

    def "ignores variant dimension with only one value"() {

    int dim1 = 1 << 16;
    int dim2 = 1 << 15;
    assertThat(dim1 * dim2).isLessThan(0);
    testConcat_overflow(dim1, dim2);
  }

  @GwtIncompatible // different overflow behavior; could probably be made to work by using ~~
  public void testConcat_overflow_nonNegative() {
    int dim1 = 1 << 16;
    int dim2 = 1 << 16;
    assertThat(dim1 * dim2).isAtLeast(0);

    int dim1 = 1 << 16;
    int dim2 = 1 << 15;
    assertThat(dim1 * dim2).isLessThan(0);
    testConcat_overflow(dim1, dim2);
  }

  @GwtIncompatible // different overflow behavior; could probably be made to work by using ~~
  public void testConcat_overflow_nonNegative() {
    int dim1 = 1 << 16;
    int dim2 = 1 << 16;
    assertThat(dim1 * dim2).isAtLeast(0);

  }
  float& arg0(size_t dim0, size_t dim1) {
    return (*static_cast<float(*)[1][2]>(
        arg_data(0)))[dim0][dim1];
  }
  const float* arg0_data() const {
    return static_cast<const float*>(arg_data(0));
  }
  const float& arg0(size_t dim0, size_t dim1) const {
    return (*static_cast<const float(*)[1][2]>(
        arg_data(0)))[dim0][dim1];
  }
  int arg0_size() const {

// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

package math

// Dim returns the maximum of x-y or 0.
//
// Special cases are:
//
//	Dim(+Inf, +Inf) = NaN
//	Dim(-Inf, -Inf) = NaN
//	Dim(x, NaN) = Dim(NaN, x) = NaN
func Dim(x, y float64) float64 {
	// The special cases result in NaN after the subtraction:
	//      +Inf - +Inf = NaN
	//      -Inf - -Inf = NaN

    return a;
  }
  int64_t rank = a.getRank();
  SmallVector<int64_t, 4> dims;
  dims.resize(rank);
  for (int i = 0, e = rank; i != e; i++) {
    int64_t dim0 = a.getDimSize(i);
    int64_t dim1 = b.getDimSize(i);
    dims[i] = (dim0 == ShapedType::kDynamic) ? dim1 : dim0;
  }
  return RankedTensorType::get(dims, a.getElementType());
}
}  // namespace detail

    absl::string_view node_shapes_str) {
  std::vector<int> dims;
  for (absl::string_view dim_str : absl::StrSplit(node_shapes_str, ',')) {
    // Treats empty input shape as scalar
    if (dim_str.empty()) continue;
    if (dim_str == "?") {
      dims.push_back(-1);
      continue;
    }
    int size;
    TF_RET_CHECK(absl::SimpleAtoi(dim_str, &size));
    dims.push_back(size);
  }
  return dims;
}

    auto dim_nums = first_dot.getDotDimensionNumbers();
    return llvm::is_contained(dim_nums.getLhsBatchingDimensions(), dim) ||
           llvm::is_contained(dim_nums.getLhsContractingDimensions(), dim);
  };

  // dot_general outputs are always in the
  //   [batch dims, LHS other dims, RHS other dims]
  // layout, so the new concat dim is where the n-th (base-0 counting) LHS other

namespace tensorflow {
namespace {

std::string ConvertToMlirString(const std::vector<int64_t>& dims,
                                bool unknown_rank, DataType dtype) {
  TensorShapeProto shape;
  shape.set_unknown_rank(unknown_rank);
  for (int64_t dim : dims) {
    shape.add_dim()->set_size(dim);
  }
  mlir::MLIRContext context;
  mlir::Builder b(&context);

    // Allocate output
    TF_Status* s = TF_NewStatus();
    int64_t dim = 1;
    size_t tensor_size_bytes = TF_DataTypeSize(TF_FLOAT);
    TF_Tensor* output = TF_AllocateOutput(
        /*context=*/ctx, /*index=*/0, /*dtype=*/TF_FLOAT, /*dims=*/&dim,
        /*num_dims=*/1, /*len=*/tensor_size_bytes, s);
    validate_tensor(output, &dim, 1, TF_FLOAT);

    // Set output to 3
    float values[1] = {3.0f};