final inputFiles = [file: [inputFile], dir: [inputDir], missingFile: [missingInputFile]]
    final outputFiles = [file: outputFile, missingFile: missingOutputFile]
    final outputDirs = [emptyDir: emptyOutputDir, dir: outputDir, missingDir: missingOutputDir]
    final createFiles = [outputFile, outputDirFile, outputDirFile2] as Set

    def unitOfWork = builder.build()

        k8s.io/kube-openapi/cmd/openapi-gen

    # The result file, in each pkg, of open-api generation.
    local output_file="${GENERATED_FILE_PREFIX}openapi.go"

    local output_dir="pkg/generated/openapi"
    local output_pkg="k8s.io/kubernetes/${output_dir}"
    local known_violations_file="${API_KNOWN_VIOLATIONS_DIR}/violation_exceptions.list"

    local report_file="${OUT_DIR}/api_violations.report"

      int64_t output_dim = output_shape[spatial_dim_index];
      if (!static_dims(input_dim, output_dim)) return success();

      int64_t input_dim_pad = input_dim * block_size;
      // If crops are unknown, the maximum output spatial dim size is input
      // spatial dim size * block_size, as crops can be minimum 0.
      if (crops_values.empty() && output_dim > input_dim * block_size)

    compat_api_versions = [
        1,
        2,
    ],
    compat_init_templates = [
        "compat_template_v1.__init__.py",
        "compat_template.__init__.py",
    ],
    output_dir = "_api/v1/",
    output_files = TENSORFLOW_API_INIT_FILES_V1,
    output_package = "tensorflow._api.v1",
    packages_to_ignore = ["tensorflow.python.framework.test_ops"],
    root_file_name = "v1.py",

init-ztunnel-rs:
	TARGET_OUT=$(TARGET_OUT) bin/build_ztunnel.sh

# Pull dependencies such as envoy
depend: init | $(TARGET_OUT)

DIRS_TO_CLEAN := $(TARGET_OUT)
DIRS_TO_CLEAN += $(TARGET_OUT_LINUX)

$(OUTPUT_DIRS):
	@mkdir -p $@

.PHONY: ${GEN_CERT}
GEN_CERT := ${ISTIO_BIN}/generate_cert
${GEN_CERT}:
	GOOS=$(GOOS_LOCAL) && GOARCH=$(GOARCH_LOCAL) && common/scripts/gobuild.sh $@ ./security/tools/generate_cert

          (*should_revisit)[e->dst()->id()] = true;
        }
      }
    }
  }

  void SetPredicate(Node* n, absl::Span<const int> output_idxs, Predicate* pred,
                    std::vector<bool>* should_revisit) {
    for (int output_idx : output_idxs) {
      SetPredicate(n, output_idx, pred, should_revisit);
    }
  }

  Status HandleSwitch(Node* n, std::vector<bool>* should_revisit);

                @InputFiles def inputFiles = project.layout.files()
                @OutputFiles Map<String, File> outputFiles = [:]
                @OutputDirectories Map<String, File> outputDirs = [:]
                @TaskAction void action() {}
            }
        """

        buildFile << """
            task test(type: TaskWithValidOutputFilesAndOutputDirectoriesProperty) {
            }

        if (!ShapedType::isDynamic(input_dim) &&
            !ShapedType::isDynamic(output_dim) && output_dim != input_dim * m) {
          return op.emitOpError()
                 << "requires input.shape[" << i << "] (" << input_dim << ")"
                 << " * " << m << " to be equal to "
                 << "output.shape[" << i << "] (" << output_dim << ")";
        }
      }
    }
  }

  return success();

  let summary = "Computes gradients for SparseSegmentMean.";

  let description = [{
Returns tensor "output" with same shape as grad, except for dimension 0 whose
value is output_dim0.
  }];

  let arguments = (ins
    Arg<TF_FloatTensor, [{gradient propagated to the SparseSegmentMean op.}]>:$grad,