}

  TF_Operation* UseT(const std::vector<TF_Output>& inputs) {
    TF_Operation* op;
    UseHelper(inputs, &op);
    return op;
  }

  // All the *Helper methods are used as a workaround for the restrictions that
  // one cannot call ASSERT_* methods in non-void-returning functions (when
  // exceptions are disabled during compilation)
  void UseHelper(const std::vector<TF_Output>& inputs, TF_Operation** op) {

* For some kinds of configuration inputs, it is possible to use the opt-out options that make Gradle fall back to the earlier behavior, omitting the inputs from detection.
This temporary workaround is aimed to mitigate performance issues coming from out-of-date plugins.

It is possible to temporarily opt out of configuration input detection in the following cases:

      // `element_shape` to 1 here.
      //
      // This workaround enables converting Keras RNN without specifying batch
      // dimension. This isn't guaranteed to work, but it doesn't break any
      // non-broken cases either (since it's already broken if `element_shape`
      // contains -1).
      // TODO(b/142096690): Support dynamic element shape and remove the
      // workaround.
      SmallVector<int32_t, 4> new_element_shape_values;

[[kotlin_dsl_plugins_catalogs_workaround]]
==== Workaround for false positive errors shown in Kotlin DSL `plugins {}` block using version catalog is not needed anymore

Version catalog accessors for plugin aliases in the `plugins {}` block aren't shown as errors in IntelliJ IDEA and Android Studio Kotlin script editor anymore.

If you were using the `@Suppress("DSL_SCOPE_VIOLATION")` annotation as a workaround, you can now remove it.

# https://github.com/tensorflow/tensorflow/blob/cd67f4f3723f9165aabedd0171aaadc6290636e5/tensorflow/tensorflow.bzl#L396-L425
# And is usable in the "deps" attribute instead of the "srcs" attribute
# as a workaround for https://github.com/tensorflow/tensorflow/issues/34117
cc_import(
    name = "libtensorflow_framework_import_lib",
    shared_library = select({
        "//tensorflow:macos": ":libtensorflow_framework.dylib",

            // The transformation may need to access dependencies raw models,
            // which may cause some re-entrance in the build() method and can
            // actually cause deadlocks.  In order to workaround the problem,
            // we do a first pass by reading all rawModels in order.
            List<ProjectBuildingResult> results = new ArrayList<>();
            boolean failure = false;

        def property = propertyWithNoValue()
        def function = Mock(Callable)
        def provider = new DefaultProvider<T>(function)
        property.set(provider)

        // TODO - this verifies that a workaround to accidental concurrent access works. Instead, this should really be an error

        when:
        property.finalizeValue()

        then:
        1 * function.call() >> {

    if (target_opset_ == OpSet::UNIFORM_QUANTIZED) {
      new_type = ConvertIntToQint(new_type, rewriter.getContext());

      // TODO(b/225793355): It adds TensorProtoAttr to the constant as a
      // workaround.
      tensorflow::TensorProto tensor_proto;
      if (!mlir::tfg::ConvertToTensorProto(
               mlir::cast<ElementsAttr>(tensor_proto_attr), &tensor_proto)
               .ok()) {
        return failure();

 * publicly visible subclass), and I doubt any of our current or future Kotlin extensions for the
 * package will refer to the class. Plus, @ParametricNullness is only a temporary workaround,
 * anyway, so we just need to get by without the annotations here until Kotlin better understands
 * our other nullness annotations.
 */
final class Synchronized {
  private Synchronized() {}

 * publicly visible subclass), and I doubt any of our current or future Kotlin extensions for the
 * package will refer to the class. Plus, @ParametricNullness is only a temporary workaround,
 * anyway, so we just need to get by without the annotations here until Kotlin better understands
 * our other nullness annotations.
 */
final class Synchronized {
  private Synchronized() {}