return stringWriter.toString();
  }

  /**
   * Returns the stack trace of {@code throwable}, possibly providing slower iteration over the full
   * trace but faster iteration over parts of the trace. Here, "slower" and "faster" are defined in
   * comparison to the normal way to access the stack trace, {@link Throwable#getStackTrace()

        # Make the FIRST file slower *deterministically*
        if self.filename == "slow.txt":
            await anyio.sleep(0.05)
        return await original_read(self, size)

    monkeypatch.setattr(StarletteUploadFile, "read", patched_read)

    client = TestClient(app)

    files = [
        ("files", ("slow.txt", b"A" * 10, "text/plain")),

          values:
            - "windows-2012-r2"
            - "windows-2016"
            - "windows-2019"
      # We shred out Windows testing into 4 parallel builds like on intake for expediency.
      # Our tests run much slower on Windows so this avoids issues with builds timing out.
      - axis:
          type: user-defined
          name: GRADLE_TASK
          values:
            - 'checkPart1'
            - 'checkPart2'

  }

  /**
   * Similar to load64, but allows offset + 8 > input.length, padding the result with zeroes. This
   * has to explicitly reverse the order of the bytes as it packs them into the result which makes
   * it slower than the native version.
   *
   * @param input the input bytes
   * @param offset the offset into the array at which to start reading
   * @param length the number of bytes from the input to read

### 3. Failure after conversion
If the conversion is successful, but the generated model is wrong, then state what is wrong:

- Model produces wrong results and/or has lesser accuracy.
- Model produces correct results, but it is slower than expected.

### 4. (optional) RNN conversion support
If converting TF RNN to TFLite fused RNN ops, please prefix [RNN] in the title.

### 5. (optional) Any other info / logs

Java serialization is one way to implement this, however, despite its simplicity of implementation, it has several drawbacks:

- **Performance:**
Java's built-in serialization mechanism is often slower compared to other serialization solutions.
This is due to Java's use of reflection and the need to maintain a lot of metadata.

- **Size of Serialized Data:**

    return stringWriter.toString();
  }

  /**
   * Returns the stack trace of {@code throwable}, possibly providing slower iteration over the full
   * trace but faster iteration over parts of the trace. Here, "slower" and "faster" are defined in
   * comparison to the normal way to access the stack trace, {@link Throwable#getStackTrace()

* `compresslevel` - Used during GZip compression. It is an integer ranging from 1 to 9. Defaults to `9`. Lower value results in faster compression but larger file sizes, while higher value results in slower compression but smaller file sizes.

## Other middlewares { #other-middlewares }

There are many other ASGI middlewares.

For example:

* **Starlette**:
    * Will have the next best performance, after Uvicorn. In fact, Starlette uses Uvicorn to run. So, it probably can only get "slower" than Uvicorn by having to execute more code.
    * But it provides you the tools to build simple web applications, with routing based on paths, etc.

When using a `response_model` or return type, FastAPI won't use the `jsonable_encoder` to convert the data (which would be slower) nor the `JSONResponse` class.

Instead it takes the JSON bytes generated with Pydantic using the response model (or return type) and returns a `Response` with the right media type for JSON directly (`application/json`).