validations:
      required: true

  - type: markdown
    attributes:
      value: >
        Did you *actually* encounter the need for this enhancement in a real-world scenario, or does
        it just seem like a sensible behavior for the feature to have?


        Before we make significant changes to existing features in Guava, we really want to be sure

## How to access to files inside a ZIP archive

Accessing to contents inside an archive can be done using regular S3 API with a modified request path. You just need to append the path of the content inside the archive to the path of the archive itself.

e.g.:

	needCSize := f.CompressedSize == ^uint32(0)
	needHeaderOffset := f.headerOffset == int64(^uint32(0))

	// Best effort to find what we need.
	// Other zip authors might not even follow the basic format,
	// and we'll just ignore the Extra content in that case.
	var modified time.Time
parseExtras:
	for extra := readBuf(f.Extra); len(extra) >= 4; { // need at least tag and size
		fieldTag := extra.uint16()
		fieldSize := int(extra.uint16())

    return new FunctionToCacheLoader<>(function);
  }

  /**
   * Returns a cache loader based on an <i>existing</i> supplier instance. Note that there's no need
   * to create a <i>new</i> supplier just to pass it in here; just subclass {@code CacheLoader} and
   * implement {@link #load load} instead.
   *
   * <p>The returned object is serializable if {@code supplier} is serializable.
   *

            // Do nothing - just for testing
        }

        @Override
        protected void doPost(HttpServletRequest req, HttpServletResponse resp) throws ServletException, IOException {
            // Do nothing - just for testing
        }
    }

    private TestNtlmServlet ntlmServlet;

    @Mock

#### Framed protocols

Framed protocols like http/2 don't lend themselves to blocking APIs. Each application-layer thread wants to do blocking I/O for a specific stream, but the streams are multiplexed on the socket. You can't just talk to the socket, you need to cooperate with the other application-layer threads that you're sharing it with.

{* ../../docs_src/body/tutorial001_py310.py hl[5:9] *}


The same as when declaring query parameters, when a model attribute has a default value, it is not required. Otherwise, it is required. Use `None` to make it just optional.

For example, this model above declares a JSON "`object`" (or Python `dict`) like:

```JSON
{
    "name": "Foo",
    "description": "An optional description",
    "price": 45.2,
    "tax": 3.5
}

import org.jspecify.annotations.NullUnmarked;

/**
 * A {@link Future} implementation which always throws directly from calls to {@code get()} (i.e.
 * not wrapped in {@code ExecutionException}. For just a normal failure, use {@link
 * SettableFuture}).
 *
 * <p>Useful for testing the behavior of Future utilities against odd futures.
 *
 * @author Anthony Zana
 */
@GwtCompatible
@NullUnmarked

  public HashCode hash() {
    return hasher.hash();
  }

  // Overriding close() because FilterOutputStream's close() method pre-JDK8 has bad behavior:
  // it silently ignores any exception thrown by flush(). Instead, just close the delegate stream.
  // It should flush itself if necessary.
  @Override
  public void close() throws IOException {
    out.close();
  }

        ArtifactResolutionResult result = resolve(request);

        // We have collected all the problems so let's mimic the way the old code worked and just blow up right here.
        // That's right lets just let it rip right here and send a big incomprehensible blob of text at unsuspecting
        // users. Bad dog!

        resolutionErrorHandler.throwErrors(request, result);