# shellcheck disable=SC2039,SC3045
        ulimit -n "$MAX_FD" ||
            warn "Could not set maximum file descriptor limit to $MAX_FD"
    esac
fi

# Collect all arguments for the java command, stacking in reverse order:
#   * args from the command line
#   * the main class name
#   * -classpath
#   * -D...appname settings
#   * --module-path (only if needed)

app.add_middleware(MiddlewareB)
```

This results in the following execution order:

* **Request**: MiddlewareB → MiddlewareA → route

* **Response**: route → MiddlewareA → MiddlewareB

This stacking behavior ensures that middlewares are executed in a predictable and controllable order.

## Other middlewares { #other-middlewares }

     */
    @Nullable
    Path getPath();

    /**
     * Creates a new input stream to read the source contents.
     * Each call creates a fresh stream starting from the beginning.
     * The caller is responsible for closing the returned stream.
     *
     * @return a new input stream positioned at the start of the content

     */
    public String[] lang;

    /**
     * Array of additional query strings to exclude from search.
     */
    public String[] ex_q;

    /**
     * The starting position for search results pagination.
     */
    @ValidateTypeFailure
    public Integer start;

    /**
     * The offset for search results.
     */
    @ValidateTypeFailure
    public Integer offset;

   *
   * <ul>
   *   <li>{@linkplain LoadingCache#get(Object) awaiting the result of a pending load} rather than
   *       starting a redundant one
   *   <li>eliminating the error-prone caching boilerplate
   *   <li>tracking load {@linkplain #stats statistics}
   * </ul>
   *
   * <p>Among the further improvements that {@code LoadingCache} can provide but this method cannot:
   *
   * <ul>

    }

    /**
     * Writes a 16-bit integer value to a byte array in little-endian format
     * @param val the value to write
     * @param dst the destination byte array
     * @param dstIndex the starting index in the destination array
     */
    public static void writeInt2(final long val, final byte[] dst, int dstIndex) {
        dst[dstIndex] = (byte) val;
        dstIndex++;
        dst[dstIndex] = (byte) (val >> 8);

   *
   * <ul>
   *   <li>{@linkplain LoadingCache#get(Object) awaiting the result of a pending load} rather than
   *       starting a redundant one
   *   <li>eliminating the error-prone caching boilerplate
   *   <li>tracking load {@linkplain #stats statistics}
   * </ul>
   *
   * <p>Among the further improvements that {@code LoadingCache} can provide but this method cannot:
   *
   * <ul>

[2020-01-01 00:00:00] Q10000 starting              : OkHttp ConnectionPool
[2020-01-01 00:00:00] Q10000 run again after 300 s : OkHttp ConnectionPool
[2020-01-01 00:00:00] Q10000 finished run in   1 ms: OkHttp ConnectionPool
[2020-01-01 00:00:00] Q10001 scheduled after   0 µs: OkHttp squareup.com applyAndAckSettings
[2020-01-01 00:00:00] Q10001 starting              : OkHttp squareup.com applyAndAckSettings

 * recommended that the caller does not persist a reference to the backing map (unless the backing
 * map is immutable).
 *
 * <p>This class is tailored toward use cases in common.graph. It is *NOT* a general purpose map.
 *
 * @author James Sexton
 */
class MapIteratorCache<K, V> {
  private final Map<K, V> backingMap;

  /*

        }
    }

    /**
     * Decodes a string from UTF-16LE (Unicode Little Endian) bytes.
     *
     * @param src the byte array containing the encoded string
     * @param srcIndex the starting offset in the byte array
     * @param len the number of bytes to decode
     * @return decoded string
     */
    public static String fromUNIBytes(final byte[] src, final int srcIndex, final int len) {