return t.GetWithCtx(context.Background())
}

func (t *Cache[T]) update(ctx context.Context) error {
	val, err := t.updateFn(ctx)
	if err != nil {
		if t.opts.ReturnLastGood && t.val.Load() != nil {
			// Keep last good value, so update
			// does not return an error.
			return nil
		}
		return err
	}

	t.val.Store(&val)
	t.lastUpdateMs.Store(time.Now().UnixMilli())
	return nil

But when checking benchmarks and comparisons you should keep the following in mind.

## Benchmarks and speed { #benchmarks-and-speed }

When you check the benchmarks, it is common to see several tools of different types compared as equivalent.

    if (userRequest.header("Host") == null) {
      requestBuilder.header("Host", userRequest.url.toHostHeader())
    }

    if (userRequest.header("Connection") == null) {
      requestBuilder.header("Connection", "Keep-Alive")
    }

    // If we add an "Accept-Encoding: gzip" header field we're responsible for also decompressing
    // the transfer stream.
    var transparentGzip = false

            // xapth.<field>=<value>
        }

        // config.*
        public static class Config {
            public static final String KEEP_ORIGINAL_BODY = "keep.original.body";
            public static final String CLEANUP_ALL = "cleanup.all";
            public static final String CLEANUP_URL_FILTERS = "cleanup.urlFilters";
            public static final String JCIFS_PREFIX = "jcifs.";

type Sub[T Maskable] struct {
	ch     chan T
	types  Mask
	filter func(entry T) bool
}

// PubSub holds publishers and subscribers
type PubSub[T Maskable, M Maskable] struct {
	// atomics, keep at top:
	types          uint64
	numSubscribers int32
	maxSubscribers int32

	// not atomics:
	subs []*Sub[T]
	sync.RWMutex
}

// Publish message to the subscribers.

                }
            } catch (final Exception e) {
                // Ignore errors - some servers don't properly return this field
                // Keep the default empty string value
            }
        }

        return bufferIndex - start;
    }

    @Override
    public String toString() {

  rm -f ${tmp_png_prefix}*png
elif [[ x"${cmd_type}" = "xsvg" ]] ; then
  check_command rsvg-convert
  target_file=$(echo "$url" | sed -e "s#^file:/*#/#g")
  rsvg-convert -w 100 -h 100 --keep-aspect-ratio "${target_file}" -o "${output_file}"
elif [[ x"${cmd_type}" = "ximage" ]] ; then
  im_cmd=$(get_imagemagick_cmd)
  if [[ -z "${im_cmd}" ]] ; then
    echo "ImageMagick (convert or magick) does not exist."

if [ "x$FESS_HEAP_NEWSIZE" != "x" ]; then
    FESS_JAVA_OPTS="$FESS_JAVA_OPTS -Xmn${FESS_HEAP_NEWSIZE}"
fi

# set to headless, just in case
FESS_JAVA_OPTS="$FESS_JAVA_OPTS -Djava.awt.headless=true"

# maximum # keep-alive connections to maintain at once
FESS_JAVA_OPTS="$FESS_JAVA_OPTS -Dhttp.maxConnections=20"

# Force the JVM to use IPv4 stack
if [ "x$FESS_USE_IPV4" != "x" ]; then

     *  XPath namespace prefixes are resolved from the namespaceNode.
     *  The implementation of this is a little slow, since it creates
     *  a number of objects each time it is called.  This could be optimized
     *  to keep the same objects around, but then thread-safety issues would arise.
     *
     *  @param contextNode The node to start searching from.
     *  @param expression A valid XPath string.

        a full-time Guava team member. [Feedback](https://stackoverflow.com/a/4543114) from our
        users indicates that they really appreciate Guava's high power-to-weight ratio. It's
        important to us to keep Guava as easy to use and understand as we can. That means boiling
        features down to compact but powerful abstractions, and controlling feature bloat carefully.

  - type: textarea
    attributes:
      label: API(s)