* inputs, regardless of how the elements are divided.
   */
  @SafeVarargs
  @CanIgnoreReturnValue
  @SuppressWarnings("nullness") // TODO(cpovirk): Remove after we fix whatever the bug is.
  public final CollectorTester<T, A, R> expectCollects(R expectedResult, T... inputs) {
    List<T> list = Arrays.asList(inputs);
    doExpectCollects(expectedResult, list);

     * that seems unlikely enough to be worth complicating the test over, especially if there's any
     * chance that a permissive test could mask a bug.
     */
    expectUnchanged();
    // Be extra thorough in case internal state was corrupted by the expected null.
    assertEquals(new ArrayList<>(), new ArrayList<>(multimap().get(k3())));

func (n *NetworkError) Unwrap() error {
	return n.Err
}

// Client - http based RPC client.
type Client struct {
	connected int32 // ref: https://golang.org/pkg/sync/atomic/#pkg-note-BUG
	_         int32 // For 64 bits alignment
	lastConn  int64

	// HealthCheckFn is the function set to test for health.
	// If not set the client will not keep track of health.

          // I regard highly unlikely that both the little endianness tests above and this one
          // passes, and there is still a little endianness bug lurking around.
        }
      }

      Control control = new Control();
      Hasher controlSink = control.newHasher(1024);

      Iterable<Hasher> sinksAndControl =

        classRepository.each {name, ClassMetaData metaData ->
            linkRepository.put(name, new ClassLinkMetaData(metaData))
        }

        // workaround to IBM JDK bug
        def createDslDocModelClosure = this.&createDslDocModel.curry(classDocbookDirectory.get().asFile, mainDocbookTemplate, classRepository)

        def doc = mainDocbookTemplate
        use(BuildableDOMCategory) {

* **Uvicorn**:
    * 具有最佳性能，因为除了服务器本身外，它没有太多额外的代码。
    * 您不会直接在 Uvicorn 中编写应用程序。这意味着您的代码至少必须包含 Starlette（或 **FastAPI**）提供的代码。如果您这样做了（即直接在 Uvicorn 中编写应用程序），最终的应用程序会和使用了框架并且最小化了应用代码和 bug 的情况具有相同的性能损耗。
    * 如果要对比与 Uvicorn 对标的服务器，请将其与 Daphne，Hypercorn，uWSGI等应用服务器进行比较。
* **Starlette**:
    * 在 Uvicorn 后使用 Starlette，性能会略有下降。实际上，Starlette 使用 Uvicorn运行。因此，由于必须执行更多的代码，它只会比 Uvicorn 更慢。

* maximize the chance of your changes being accepted
* work on the Gradle code base
* get help if you encounter trouble

## Before you start

Before starting to work on a feature or a bug fix, please open an issue to discuss the use case or bug with us, or post a comment in the relevant issue. 
This can save everyone a lot of time and frustration.

For any non-trivial change, we need to be able to answer these questions:

							<goal>check</goal>
						</goals>
					</execution>
				</executions>
				<configuration>
					<failOnError>false</failOnError>
					<ignored>
						<difference>
							<!-- seems to be a clirr bug that it cannot handle the different inherited 
								return types -->
							<className>**</className>
							<differenceType>7006</differenceType>
							<method>*</method>
							<from>*</from>
							<to>*</to>

```


Advanced Profiling
------------------

Android Studio’s Advanced Profiling feature rewrites OkHttp bytecode for instrumentation.
Unfortunately it crashes on OkHttp 4.x’s bytecode. Until [Google’s bug][advanced_profiling_bug] is
fixed you must disable advanced profiling in Android Studio.

![Disable Advanced Profiling](../assets/images/******@****.***)


R8 / ProGuard
-------------

    hashed_password = hashed_password,
)
```

/// warning | 警告

辅助的附加函数只是为了演示可能的数据流，但它们显然不能提供任何真正的安全机制。

///

## 减少重复

**FastAPI** 的核心思想就是减少代码重复。

代码重复会导致 bug、安全问题、代码失步等问题（更新了某个位置的代码，但没有同步更新其它位置的代码）。

上面的这些模型共享了大量数据，拥有重复的属性名和类型。

FastAPI 可以做得更好。

声明 `UserBase` 模型作为其它模型的基类。然后，用该类衍生出继承其属性（类型声明、验证等）的子类。

所有数据转换、校验、文档等功能仍将正常运行。