],
    defines = if_llvm_x86_available(["TF_LLVM_X86_AVAILABLE=1"]),
    deps = [
        ":tfcompile_lib",
        "//tensorflow/core:lib",
        "//tensorflow/core:protos_all_cc",
        "//tensorflow/core:test",
        "//tensorflow/core:test_main",
        "//tensorflow/core/platform:resource_loader",
        "@com_google_absl//absl/strings",
        "@llvm-project//llvm:Support",  # fixdeps: keep

// attached as an attribute.
std::unique_ptr<OperationPass<func::FuncOp>>
CreateMaterializePassthroughOpPass();

// Replicates the TensorList init op by undoing some CSE needed for correct
// shape assignment in shape_inference.
std::unique_ptr<OperationPass<func::FuncOp>>
CreateReplicateTensorListInitOpsPass();

// Performs Shape Inference on the TensorFlow dialect using the global registry.

    // Currently flaky with CC enabled
    def "producer task followed by a destroyer task followed by a producer with a dependency on the first producer are run in the correct order"() {
        def foo = subproject(':foo')
        def bar = subproject(':bar')

        def cleanFooLocal = foo.task('cleanFooLocalState').destroys('build/foo-local')

    if (dividend >= 0) {
      return dividend / divisor;
    }

    /*
     * Otherwise, approximate the quotient, check, and correct if necessary. Our approximation is
     * guaranteed to be either exact or one less than the correct value. This follows from fact that
     * floor(floor(x)/i) == floor(x/i) for any real x and integer i != 0. The proof is not quite
     * trivial.
     */

    if (dividend >= 0) {
      return dividend / divisor;
    }

    /*
     * Otherwise, approximate the quotient, check, and correct if necessary. Our approximation is
     * guaranteed to be either exact or one less than the correct value. This follows from fact that
     * floor(floor(x)/i) == floor(x/i) for any real x and integer i != 0. The proof is not quite
     * trivial.
     */

	mask := ^uint64(0) >> (64 - n)

	// Calculate the multiplicative inverse via Newton's method.
	// Quadratic convergence doubles the number of correct bits per iteration.
	m := d0            // initial guess correct to 3-bits d0*d0 mod 8 == 1
	m = m * (2 - m*d0) // 6-bits
	m = m * (2 - m*d0) // 12-bits
	m = m * (2 - m*d0) // 24-bits
	m = m * (2 - m*d0) // 48-bits

        ":test_graph_tftop_k",
        ":test_graph_tfvariable",
        ":test_graph_tfvariable_readonly",
        ":test_graph_tfvariable_sequential_updates",
        "//tensorflow/core:lib",
        "//tensorflow/core:test",
        "//tensorflow/core:test_main",
        "//tensorflow/core/platform:regexp",
        "@com_google_absl//absl/strings",
        "@eigen_archive//:eigen3",
        "@local_xla//xla:shape_util",

If you added the return type annotation, tools and editors would complain with a (correct) error telling you that your function is returning a type (e.g. a dict) that is different from what you declared (e.g. a Pydantic model).

    //If you're changing this you will also need to change: src/snippets/modelRules/basicRuleSourcePlugin/basicRuleSourcePlugin-model-task.out
    def "displays a report in the correct format"() {
        given:
        settingsFile << "rootProject.name = 'test'"
        buildFile << """

@Managed
public interface PasswordCredentials {
    String getUsername()
    String getPassword()

                    System.out.println("afterClass out"); System.err.println("afterClass err");
                }
            }
        """
    }

    def "attaches events to correct test descriptors of a suite"() {
        buildFile << "test.useTestNG { suites 'suite.xml' }"

        file("suite.xml") << """<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE suite SYSTEM "http://testng.org/testng-1.0.dtd">