So, you could add additional data to the automatically generated schema.

For example, you could decide to read and validate the request with your own code, without using the automatic features of FastAPI with Pydantic, but you could still want to define the request in the OpenAPI schema.

You could do that with `openapi_extra`:

{* ../../docs_src/stream_data/tutorial001_py310.py ln[32:35] hl[33] *}

This also means that with `StreamingResponse` you have the **freedom** and **responsibility** to produce and encode the data bytes exactly as you need them to be sent, independent of the type annotations. 🤓

### Stream Bytes { #stream-bytes }

One of the main use cases would be to stream `bytes` instead of strings, you can of course do it.

 *
 * The first element on each line is a single hex code point (like 0041) or a hex code point range
 * (like 0030..0039).
 *
 * The second element on each line is a mapping type, like `valid` or `mapped`.
 *
 * For lines that contain a mapping target, the next thing is a sequence of hex code points (like
 * 0031 2044 0034).
 *
 * All other data is ignored.
 */

Guava, the best approach is to create an [issue][] (or comment on an existing
issue if there is one) requesting the feature and describing specific use cases
for it.

If we decide to pursue a feature request, it will go through a thorough process
of API design and review. Any code should come after this.

[APIs]: https://en.wikipedia.org/wiki/API
[issue]: https://github.com/google/guava/issues

Pull requests
-------------

import java.util.Map;

import javax.security.auth.kerberos.KerberosKey;

import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import jcifs.util.Hexdump;

/**
 * Privilege Attribute Certificate (PAC) decoder for Kerberos tickets.
 * Parses and validates PAC data structures containing user authorization information
 * from Active Directory Kerberos tickets.
 */
public class Pac {

- `compresslevel` - GZip 壓縮時使用的等級。為 1 到 9 的整數。預設為 `9`。值越小壓縮越快但檔案較大，值越大壓縮較慢但檔案較小。

## 其他中介軟體 { #other-middlewares }

還有許多其他 ASGI 中介軟體。

例如：

- [Uvicorn 的 `ProxyHeadersMiddleware`](https://github.com/encode/uvicorn/blob/master/uvicorn/middleware/proxy_headers.py)
- [MessagePack](https://github.com/florimondmanca/msgpack-asgi)

					return
				}
			case <-ctx.Done():
				return
			}

			if !tickered {
				h.totalMessages.Add(1)
				if !isDirQueue {
					if err := enc.Encode(&entry); err != nil {
						h.config.LogOnceIf(
							ctx,
							fmt.Errorf("unable to encode webhook log entry, err  '%w' entry: %v\n", err, entry),
							h.Name(),
						)
						h.failedMessages.Add(1)
						continue
					}
				} else {

		for i, disk := range disks {
			writers[i] = newBitrotWriter(disk, "", "testbucket", "testobject", erasure.ShardFileSize(test.size), test.algorithm, erasure.ShardSize())
		}
		_, err = erasure.Encode(t.Context(), bytes.NewReader(data), writers, buffer, erasure.dataBlocks+1)
		closeBitrotWriters(writers)
		if err != nil {
			t.Fatalf("Test %d: failed to create random test data: %v", i, err)
		}

## Version 1.5.2

_2014-03-17_

 * Fix bug where deleting a file that was absent from the `HttpResponseCache`
   caused an IOException.
 * Fix bug in HTTP/2 where our HPACK decoder wasn't emitting entries in
   certain eviction scenarios, leading to dropped response headers.

## Version 1.5.1

_2014-03-11_

 * Fix 1.5.0 regression where connections should not have been recycled.

    sepStopwatchA.set(null);
    GcFinalization.awaitClear(sepStopwatchRef.get());
    // Return a weak reference to the parallel ClassLoader. This is the reference that should
    // eventually become clear if there are no other references to the ClassLoader.
    return new WeakReference<ClassLoader>(sepLoader);
  }

  /**
   * Tests that the use of a {@link FinalizableReferenceQueue} does not subsequently prevent the