if (UPDATE_BB != null) {
        try {
          UPDATE_BB.invokeExact(cs, bb);
        } catch (Throwable e) {
          throwIfUnchecked(e);
          // This should be impossible, since `update` has no `throws` clause.
          throw new UndeclaredThrowableException(e);
        }
        return true;
      } else {
        return false;
      }
    }

// GraphDef type. These functions must be deleted by calling TF_DeleteFunction.
//
// If `mutate_proto_func` is non-NULL, run it over each FunctionDef proto,
// before creating a TF_Function out of the possibly mutated proto.
static std::vector<UniqueFuncPtr> CreateFunctionsFromTextProto(
    const char* text_proto,
    std::function<void(FunctionDef*)>* mutate_proto_func, TF_Status* status) {
  tensorflow::GraphDef gdef;

be reported to TensorFlow’s Project Stewards, Thea Lamkin (******@****.***) and Joana Carrasqueira (******@****.***). The Project Steward will determine whether the Code of Conduct was violated, and will issue an appropriate sanction, possibly including a written warning or expulsion from the project, project sponsored spaces, or project forums. We ask that you make a good-faith effort to resolve your conflict via the conflict resolution policy before submitting a report.

Violations...

     * repository. It is worth to point out that merging does not simply choose one of the input repositories and
     * discards the others but actually combines their possibly different policies.
     *
     * @param repositories The original repositories, may be {@code null}.
     * @return The effective repositories or {@code null} if the input was {@code null}.
     */

The same way, you can define logic (code) that should be executed when the application is **shutting down**. In this case, this code will be executed **once**, **after** having handled possibly **many requests**.

Mas em algum ponto, não houve outra opção senão criar algo que fornecesse todos esses recursos, pegando as melhores idéias de ferramentas anteriores, e combinando eles da melhor forma possível, utilizando recursos da linguagem que não estavam disponíveis antes (_Type Hints_ no Python 3.6+).

## Ferramentas anteriores

### <a href="https://www.djangoproject.com/" class="external-link" target="_blank">Django</a>

import java.util.List;
import java.util.Map;
import junit.framework.TestCase;

/**
 * Unit tests of {@link TypeResolver}.
 *
 * @author Ben Yu
 */
@AndroidIncompatible // lots of failures, possibly some related to bad equals() implementations?
public class TypeResolverTest extends TestCase {

  public void testWhere_noMapping() {
    Type t = aTypeVariable();
    assertEquals(t, new TypeResolver().resolveType(t));

import java.util.List;
import java.util.Map;
import junit.framework.TestCase;

/**
 * Unit tests of {@link TypeResolver}.
 *
 * @author Ben Yu
 */
@AndroidIncompatible // lots of failures, possibly some related to bad equals() implementations?
public class TypeResolverTest extends TestCase {

  public void testWhere_noMapping() {
    Type t = aTypeVariable();
    assertEquals(t, new TypeResolver().resolveType(t));

Já é possível criar aplicativos de produção com **FastAPI** (e provavelmente você já faz isso há algum tempo), apenas precisando ter certeza de usar uma versão que funcione corretamente com o resto do seu código.

## Fixe a sua versão de `fastapi`

    boolean threadStarted = false;
    try {
      startFinalizer.invoke(null, FinalizableReference.class, queue, frqRef);
      threadStarted = true;
    } catch (IllegalAccessException impossible) {
      throw new AssertionError(impossible); // startFinalizer() is public
    } catch (Throwable t) {
      logger.log(
          Level.INFO,
          "Failed to start reference finalizer thread."