if DB.Unscoped().Model(&person).Association("Addresses").Count() != 0 {
		t.Fatalf("address should be deleted when clear with unscoped")
	}

	address2_1 := Address{Name: "address 2-1"}
	address2_2 := Address{Name: "address 2-2"}
	person2 := Person{Name: "person_2", Addresses: []Address{address2_1, address2_2}}
	DB.Create(&person2)
	if err := DB.Select(clause.Associations).Delete(&person2).Error; err != nil {

```

En algunos casos, incluso es posible tener modelos de Pydantic v1 y v2 en la misma path operation de tu app de FastAPI:

{* ../../docs_src/pydantic_v1_in_v2/tutorial003_an_py310.py hl[2:3,6,12,21:22] *}

En el ejemplo anterior, el modelo de entrada es un modelo de Pydantic v1 y el modelo de salida (definido en `response_model=ItemV2`) es un modelo de Pydantic v2.

### Parámetros de Pydantic v1 { #pydantic-v1-parameters }

    assertThat(frameLog(true, 0, 0, TYPE_SETTINGS, FLAG_ACK))
      .isEqualTo("<< 0x00000000     0 SETTINGS      ACK")
    assertThat(frameLog(true, 3, 22, TYPE_HEADERS, FLAG_END_HEADERS))
      .isEqualTo("<< 0x00000003    22 HEADERS       END_HEADERS")
    assertThat(frameLog(true, 3, 226, TYPE_DATA, FLAG_END_STREAM))
      .isEqualTo("<< 0x00000003   226 DATA          END_STREAM")
    assertThat(frameLog(false, 0, 8, TYPE_GOAWAY, FLAG_NONE))

   * `Synchronized` doesn't guarantee that it will hold the mutex for those calls because callers
   * are responsible for taking the mutex themselves:
   * https://docs.oracle.com/en/java/javase/22/docs/api/java.base/java/util/Collections.html#synchronizedCollection(java.util.Collection)
   *
   * Similarly, we avoid having those methods *implemented* in terms of *other* TestSet methods

Pour ces cas, vous pouvez spécifier un appel à `jsonable_encoder` pour convertir vos données avant de les passer à une réponse :

{* ../../docs_src/response_directly/tutorial001.py hl[6:7,21:22] *}

/// note | Détails techniques

Vous pouvez aussi utiliser `from starlette.responses import JSONResponse`.

           * com.google.common.collect.testing.AbstractIteratorTester from class
           * com.google.common.collect.MultimapsTest"
           *
           * ...when we build with JDK 22 and run under JDK 8.
           */
          || info.getName().contains("MultimapsTest")
      /*
       * Luckily, we don't care about analyzing tests at all. We'd skip them all if we could do so

의존성을 해결할 때 **FastAPI**는 이 `checker`를 다음과 같이 호출합니다:

```Python
checker(q="somequery")
```

...그리고 이때 반환되는 값을 *경로 연산 함수*의 `fixed_content_included` 매개변수로 전달합니다:

{* ../../docs_src/dependencies/tutorial011_an_py39.py hl[22] *}

/// tip | 참고

이 모든 과정이 복잡하게 느껴질 수 있습니다. 그리고 지금은 이 방법이 얼마나 유용한지 명확하지 않을 수도 있습니다.

이 예시는 의도적으로 간단하게 만들었지만, 전체 구조가 어떻게 작동하는지 보여줍니다.

보안 관련 장에서는 이와 같은 방식으로 구현된 편의 함수들이 있습니다.

```

Em alguns casos, é até possível ter modelos Pydantic v1 e v2 na mesma operação de rota na sua aplicação FastAPI:

{* ../../docs_src/pydantic_v1_in_v2/tutorial003_an_py310.py hl[2:3,6,12,21:22] *}

No exemplo acima, o modelo de entrada é um modelo Pydantic v1, e o modelo de saída (definido em `response_model=ItemV2`) é um modelo Pydantic v2.

### Parâmetros do Pydantic v1 { #pydantic-v1-parameters }

           * com.google.common.collect.testing.AbstractIteratorTester from class
           * com.google.common.collect.MultimapsTest"
           *
           * ...when we build with JDK 22 and run under JDK 8.
           */
          || info.getName().contains("MultimapsTest")
      /*
       * Luckily, we don't care about analyzing tests at all. We'd skip them all if we could do so

    }

    private
    fun getMergedContributorPullRequests(pageNumber: Int): List<GitHubPullRequest> {
        val uri = "https://api.github.com/search/issues?q=is:pr+is:merged+repo:gradle/gradle+label:%22from:contributor%22&sort=updated&order=desc&per_page=$PAGE_SIZE&page=$pageNumber"
        return invokeGitHubApi(uri, GitHubPullRequestSearchResult::class.java).items
    }