* AbstractMapBasedMultimap} calls {@link #createCollection()} to create the collection of values
 * for that key. The subclass should not call {@link #createCollection()} directly, and a new
 * instance should be created every time the method is called.
 *
 * <p>For example, the subclass could pass a {@link java.util.TreeMap} during construction, and

 * thought of as overkill; however, it's difficult to determine which proper subset of this massive
 * set would be sufficient to expose any possible bug. Brute force is simpler.
 *
 * <p>To use this class the concrete subclass must implement the {@link
 * IteratorTester#newTargetIterator()} method. This is because it's impossible to test an Iterator
 * without changing its state, so the tester needs a steady supply of fresh Iterators.
 *

 * single row key may or may not be ordered, depending on the implementation. When rows and columns
 * are both sorted, it's easier to use the {@link TreeBasedTable} subclass.
 *
 * <p>The {@link #rowKeySet} method returns a {@link SortedSet} and the {@link #rowMap} method
 * returns a {@link SortedMap}, instead of the {@link Set} and {@link Map} specified by the {@link
 * Table} interface.
 *

    }

    private static String methodSignaturesMatch(CtMethod methodInSubClass, CtMethod method) {
        return methodParametersFromSignature(methodInSubClass) == methodParametersFromSignature(method)
            // Subclass can override method return type
            && methodInSubClass.returnType.subtypeOf(method.returnType)
    }

    /**

      }

      @Override
      public boolean contains(@Nullable Object key) {
        return containsKey(key);
      }
    };
  }

  // Internal methods (package-visible, but treat as only subclass-visible)

  @Nullable V getIfCached(@Nullable Object key) {
    Entry<K, V> entry = cacheEntry; // store local reference for thread-safety

 * thought of as overkill; however, it's difficult to determine which proper subset of this massive
 * set would be sufficient to expose any possible bug. Brute force is simpler.
 *
 * <p>To use this class the concrete subclass must implement the {@link
 * IteratorTester#newTargetIterator()} method. This is because it's impossible to test an Iterator
 * without changing its state, so the tester needs a steady supply of fresh Iterators.
 *

    /*
     * Yes, we sometimes set nextInKeyBucket to null, even for this "non-terminal" entry. We don't
     * do that with a plain NonTerminalImmutableMapEntry, but we do it with the BiMap-specific
     * subclass below. That's because the Entry might be non-terminal in the key bucket but terminal
     * in the value bucket (or vice versa).
     */
    private final transient @Nullable ImmutableMapEntry<K, V> nextInKeyBucket;

// =============================================================================

// Represents either a MlirTensor or a GraphTensor.
// This base class does not expose any public methods other than to distinguish
// which subclass it actually is. The user is responsible to use the right
// type of AbstractTensor in their context (do not pass an MlirTensor to a
// GraphContext and vice-versa).
class TracingTensorHandle : public AbstractTensorHandle {

 * href="http://en.wikipedia.org/wiki/Decorator_pattern">decorator pattern</a>.
 *
 * <p>This class does not implement {@link java.util.RandomAccess}. If the delegate supports random
 * access, the {@code ForwardingList} subclass should implement the {@code RandomAccess} interface.
 *
 * <p><b>Warning:</b> The methods of {@code ForwardingList} forward <b>indiscriminately</b> to the

But if you return a `Response` directly (or any subclass, like `JSONResponse`), the data won't be automatically converted (even if you declare a `response_model`), and the documentation won't be automatically generated (for example, including the specific "media type", in the HTTP header `Content-Type` as part of the generated OpenAPI).