// the call, but they may be accessed by other threads for duplex requests.

  /** True if this call still has a request body open. */
  private var requestBodyOpen = false

  /** True if this call still has a response body open. */
  private var responseBodyOpen = false

  /** True if there are more exchanges expected for this call. */

Still, in both situations, chances are that **FastAPI** will [still be faster](index.md#performance){.internal-link target=_blank} than (or at least comparable to) your previous framework.

### Dependencies

        Thread.sleep(500L)
        println("After delay: " + connectionPool.connectionCount())
      }

      connectionPool.evictAll()
      assertEquals(0, connectionPool.connectionCount()) {
        "Still ${connectionPool.connectionCount()} connections open"
      }
    }
  }

  private fun ensureAllTaskQueuesIdle() {
    val entryTime = System.nanoTime()

    ```

=== "Python 3.8+"

    ```Python hl_lines="10"
    {!> ../../../docs_src/query_params_str_validations/tutorial002_an.py!}
    ```

Notice that the default value is still `None`, so the parameter is still optional.

!!! tip
    But it is still recommended to use the ideas above, using multiple classes, instead of these parameters.

    This is because the JSON Schema generated in your app's OpenAPI (and the docs) will still be the one for the complete model, even if you use `response_model_include` or `response_model_exclude` to omit some attributes.

```Python
say_hi()  # Oh, no, this throws an error! 😱
```

The `name` parameter is **still required** (not *optional*) because it doesn't have a default value. Still, `name` accepts `None` as the value:

```Python
say_hi(name=None)  # This works, None is valid 🎉
```

It had the best performance benchmarks at the time (only surpassed by Starlette).

At first, it didn't have an automatic API documentation web UI, but I knew I could add Swagger UI to it.

It had a dependency injection system. It required pre-registration of components, as other tools discussed above. But still, it was a great feature.

    HTTP/2 to 16 MiB of unacknowledged data per connection. With this fix there is a limit of 16 MiB
    of unacknowledged data per stream and no per-connection limit.

 *  Fix: Don't close a `Deflater` while we're still using it to compress a web socket message. We
    had a severe bug where web sockets were closed on the wrong thread, which caused
    `NullPointerException` crashes in `Deflater`.

### Bigger Errors - Crashes

Nevertheless, there might be cases where we write some code that **crashes the entire application** making Uvicorn and Python crash. 💥

And still, you would probably not want the application to stay dead because there was an error in one place, you probably want it to **continue running** at least for the *path operations* that are not broken.

### Restart After Crash

### Preprocess the OpenAPI Specification for the Client Generator

The generated code still has some **duplicated information**.

We already know that this method is related to the **items** because that word is in the `ItemsService` (taken from the tag), but we still have the tag name prefixed in the method name too. 😕

We will probably still want to keep it for OpenAPI in general, as that will ensure that the operation IDs are **unique**.