* requested permits, by an invocation acquire(permits), are served from:
   *
   * - stored permits (if available)
   *
   * - fresh permits (for any remaining permits)
   *
   * How this works is best explained with an example:
   *
   * For a RateLimiter that produces 1 token per second, every second that goes by with the
   * RateLimiter being unused, we increase storedPermits by 1. Say we leave the RateLimiter unused

  tensorflow::mutex execution_mutex_;
  ExecutionState execution_state_ TF_GUARDED_BY(execution_mutex_) =
      ExecutionState::kIdle;
  // Tells the worker thread that there is new work.
  tensorflow::condition_variable start_execute_;
  // The worker thread notifies that work has finished.
  tensorflow::condition_variable finished_execute_;
  // Notifies a StartExecute that the previous Join has finished.
  tensorflow::condition_variable finished_join_;

				NewPutObjReader(hr),
				ObjectOptions{
					PreserveETag: part.ETag, // Preserve original ETag to ensure same metadata.
					IndexCB: func() []byte {
						return part.Index // Preserve part Index to ensure decompression works.
					},
					NoAuditLog: true,
				})
			if err != nil {
				return fmt.Errorf("rebalanceObject: PutObjectPart() %w", err)
			}
			parts[i] = CompletePart{
				ETag:       pi.ETag,

You can check it at <a href="http://127.0.0.1:8000/docs" class="external-link" target="_blank">http://127.0.0.1:8000/docs</a>:

<img src="/img/tutorial/behind-a-proxy/image01.png">

But if we access the docs UI at the "official" URL using the proxy with port `9999`, at `/api/v1/docs`, it works correctly! 🎉

   */
  @CanIgnoreReturnValue // pushed down from class to method
  @Override
  public Iterator<E> iterator() {
    return new Itr(toArray());
  }

  /** Snapshot iterator that works off copy of underlying q array. */
  private class Itr implements Iterator<E> {
    final Object[] array; // Array of all elements
    int cursor; // index of next element to return;

		}
		return nil
	}
	log.Infof("Repairing pod...")

	// Fetch the pod's network namespace. This must run in the host process due to how the procfs /ns/net works.
	// This will get a network namespace ID. This ID is scoped to the network namespace we running in.
	// As such, we need to be in the host namespace: the CNI pod namespace has no relation to the users pod namespace.

This also applies to `response_model_by_alias` that works similarly.

///

//// tab | Python 3.10+

```Python hl_lines="29  35"
{!> ../../docs_src/response_model/tutorial005_py310.py!}
```

////

//// tab | Python 3.8+

  private static final int MAX_HASH_BUCKET_LENGTH = 9;

  // See CompactHashMap for a detailed description of how the following fields work. That
  // description talks about `keys`, `values`, and `entries`; here the `keys` and `values` arrays
  // are replaced by a single `elements` array but everything else works similarly.

  /**
   * The hashtable object. This can be either:
   *
   * <ul>

          // Checking subtype between declared and actual type bounds
          // adds recursive isSubtypeOf() call and feels complicated.
          // There is no contract one way or another as long as isSubtypeOf() works as expected.
          combined.addAll(asList(typeParam.getBounds()));
          if (combined.size() > 1) { // Object is implicit and only useful if it's the only bound.
            combined.remove(Object.class);
          }

      reversed[i] = addr[addr.length - i - 1];
    }
    return InetAddress.getByAddress(reversed);
  }

  /**
   * Returns a new InetAddress that is one less than the passed in address. This method works for
   * both IPv4 and IPv6 addresses.
   *
   * @param address the InetAddress to decrement
   * @return a new InetAddress that is one less than the passed in address