### Creating the `Settings` only once with `lru_cache` { #creating-the-settings-only-once-with-lru-cache }

Reading a file from disk is normally a costly (slow) operation, so you probably want to do it only once and then reuse the same settings object, instead of reading it for each request.

But every time we do:

```Python
Settings()
```

          && hashCodeInsensitiveToArgReference(
              factory, args, i, generateDummyArg(param, newFreshValueGenerator()))) {
        // If the implementation uses identityHashCode(), referential equality is
        // probably intended. So no point in using an equal-but-different factory argument.
        // We check twice to avoid confusion caused by accidental hash collision.
        equalArgs.set(i, shouldBeEqualArg);
      }
    }

      try {
        return target.contains(t);
      } catch (NullPointerException | ClassCastException e) {
        return false;
      }
    }

    @Override
    /*
     * We should probably not have implemented equals() at all, but given that we did, we can't
     * provide a better implementation than the input Collection, at least without dramatic changes

        .addEqualityGroup(ImmutableIntArray.of(1, 3))
        .addEqualityGroup(ImmutableIntArray.of(1, 2, 3))
        .testEquals();
  }

  /**
   * This is probably a weird and hacky way to test what we're really trying to test, but hey, it
   * caught a bug.
   */
  public void testTrimmed() {
    ImmutableIntArray iia = ImmutableIntArray.of(0, 1, 3);

        <version>${j2objc.version}</version>
      </dependency>
      <!--
      We moved away from using dependencyManagement for test-only deps because of https://github.com/google/guava/issues/6654.
      We could probably have resumed it after https://github.com/google/guava/pull/6664.
      But it's always weird that published poms reference test-only libraries at all, so I'm not in any rush to do so.
      -->
    </dependencies>

   *
   * 1. b/192354773 in our checker affects toArray declarations.
   *
   * 2. `other[size] = null` is unsound. We could "fix" this by requiring callers to pass in an
   * array with a nullable element type. But probably they usually want an array with a non-nullable
   * type. That said, we could *accept* a `@Nullable T[]` (which, given that we treat arrays as

        <version>${j2objc.version}</version>
      </dependency>
      <!--
      We moved away from using dependencyManagement for test-only deps because of https://github.com/google/guava/issues/6654.
      We could probably have resumed it after https://github.com/google/guava/pull/6664.
      But it's always weird that published poms reference test-only libraries at all, so I'm not in any rush to do so.
      -->
    </dependencies>

     * accept that many elements.
     *
     * <p>This gets overridden in ImmutableSetMultimap.Builder to only trust the size of {@code
     * values} if it is a Set and therefore probably already deduplicated.
     */
    int expectedValueCollectionSize(int defaultExpectedValues, Iterable<?> values) {
      if (values instanceof Collection<?>) {
        Collection<?> collection = (Collection<?>) values;

        .addEqualityGroup(ImmutableDoubleArray.of(1, 3))
        .addEqualityGroup(ImmutableDoubleArray.of(1, 2, 3))
        .testEquals();
  }

  /**
   * This is probably a weird and hacky way to test what we're really trying to test, but hey, it
   * caught a bug.
   */
  public void testTrimmed() {
    ImmutableDoubleArray iia = ImmutableDoubleArray.of(0, 1, 3);

       * that behavior is undefined when the backing map is modified through another API. (It even
       * permits us to throw IllegalStateException. Maybe we should have done that, but we probably
       * shouldn't change now for fear of breaking people.)
       */
      return (lastKnownIndex == -1) ? unsafeNull() : value(lastKnownIndex);
    }

    @Override
    @ParametricNullness