* {@code equals}, {@code hashCode} or {@code toString} behavior of the returned function. A
   * future migration to {@code java.util.function} will not preserve this behavior.
   *
   * <p>As discussed above, prefer to use the method reference {@code Object::toString} instead,
   * though note that it is not serializable unless you explicitly make it {@link Serializable},

          toJoin = Arrays.copyOf(toJoin, expandedCapacity(toJoin.length, toJoin.length + 1));
        }
        toJoin[i++] = toString(part);
      }
      // We might not have seen the expected number of elements, as discussed above.
      if (i != toJoin.length) {
        toJoin = Arrays.copyOf(toJoin, i);
      }
      return String.join(separator, toJoin);
    }
    return join(parts.iterator());
  }

  /*

//
//	MinIO (<OS>; <ARCH>[; <MODE>][; dcos][; kubernetes][; docker][; source]) MinIO/<VERSION> MinIO/<RELEASE-TAG> MinIO/<COMMIT-ID> [MinIO/universe-<PACKAGE-NAME>] [MinIO/helm-<HELM-VERSION>]
//
// Any change here should be discussed by opening an issue at
// https://github.com/minio/minio/issues.
func getUserAgent(mode string) string {
	userAgentParts := []string{}
	// Helper function to concisely append a pair of strings to a

   * acquire(3) arrives. We serve this request out of storedPermits, and reduce that to 7.0 (how
   * this is translated to throttling time is discussed later). Immediately after, assume that an
   * acquire(10) request arriving. We serve the request partly from storedPermits, using all the
   * remaining 7.0 permits, and the remaining 3.0, we serve them by fresh permits produced by the

  // For a (long) discussion about this specific issue and the general futility of life.
  //
  // For the time being we are OK with the problem discussed above since it requires a caller to
  // introduce a very specific kind of data-race. And given the other operations performed by these
  // methods that involve volatile read/write operations, in practice there is no issue. Also, the

       *   columnIterator.remove() would have failed above (if the user hasn't called next() since
       *   then) or rowEntry would have been initialized by next() (as discussed above).
       */
      if (requireNonNull(rowEntry).getValue().isEmpty()) {
        rowIterator.remove();
        rowEntry = null;
      }
    }
  }

  @Override

         * Future might not even be a ListenableFuture, just a plain Future. That said, similar
         * problems can exist with methods like FutureTask.done(), not to mention slow calls to
         * Thread.interrupt() (as discussed in InterruptibleTask). At the end of the day, it's
         * unlikely that cancel() will be slow, so we can probably get away with calling it while
         * holding a lock. Still, it would be nice to avoid somehow.
         */

       *   columnIterator.remove() would have failed above (if the user hasn't called next() since
       *   then) or rowEntry would have been initialized by next() (as discussed above).
       */
      if (requireNonNull(rowEntry).getValue().isEmpty()) {
        rowIterator.remove();
        rowEntry = null;
      }
    }
  }

  @Override

    } catch (UnsupportedOperationException e) {
      return;
    }

    Set<Entry<K, V>> entrySet = map.entrySet();
    Entry<K, V> originalEntry = entrySet.iterator().next();
    // Copy the Entry, as discussed in testEntrySetRemoveAll.
    Set<Entry<K, V>> entriesToRetain =
        singleton(mapEntry(originalEntry.getKey(), originalEntry.getValue()));
    if (supportsRemove) {

At first, it didn't have an automatic API documentation web UI, but I knew I could add Swagger UI to it.

It had a dependency injection system. It required pre-registration of components, as other tools discussed above. But still, it was a great feature.