But you can return a `JSONResponse` directly from your *path operations*.

It might be useful, for example, to return custom headers or cookies.

## Return a `Response`

In fact, you can return any `Response` or any sub-class of it.

/// tip

`JSONResponse` itself is a sub-class of `Response`.

///

And when you return a `Response`, **FastAPI** will pass it directly.

 * under the License.
 */
package org.apache.maven.model.building;

/**
 * Collects problems that are encountered during model building. The primary purpose of this component is to account for
 * the fact that the problem reporter has/should not have information about the calling context and hence cannot provide
 * an expressive source hint for the model problem. Instead, the source hint is configured by the model builder before

You could do that with `openapi_extra`:

```Python hl_lines="19-36  39-40"
{!../../docs_src/path_operation_advanced_configuration/tutorial006.py!}
```

import org.apache.maven.model.Model;
import org.apache.maven.model.io.ModelParseException;

/**
 * Collects problems that are encountered during model building. The primary purpose of this component is to account for
 * the fact that the problem reporter has/should not have information about the calling context and hence cannot provide
 * an expressive source hint for the model problem. Instead, the source hint is configured by the model builder before

import org.apache.maven.api.model.InputLocation;
import org.apache.maven.api.model.Model;

/**
 * Collects problems that are encountered during model building. The primary purpose of this component is to account for
 * the fact that the problem reporter has/should not have information about the calling context and hence cannot provide
 * an expressive source hint for the model problem. Instead, the source hint is configured by the model builder before

    }

    /*
     * Otherwise, approximate the quotient, check, and correct if necessary. Our approximation is
     * guaranteed to be either exact or one less than the correct value. This follows from fact that
     * floor(floor(x)/i) == floor(x/i) for any real x and integer i != 0. The proof is not quite
     * trivial.
     */
    long quotient = ((dividend >>> 1) / divisor) << 1;

    checkNotNull(seen);
    if (!isCancelled()) {
      /*
       * requireNonNull is safe because:
       *
       * - This is a TrustedFuture, so tryInternalFastPathGetFailure will in fact return the failure
       *   cause if this Future has failed.
       *
       * - And this future *has* failed: This method is called only from handleException (through

  /**
   * Returns a {@link ScheduledExecutorService} that never executes anything.
   *
   * <p>The {@code shutdownNow} method of the returned executor always returns an empty list despite
   * the fact that everything is still technically awaiting execution. The {@code getDelay} method
   * of any {@link ScheduledFuture} returned by the executor will always return the max long value

    return FluentIterable.from(ImmutableList.<Number>of());
  }

  /*
   * The following tests demonstrate the shortcomings of or() and test that the casting workaround
   * mentioned in the method Javadoc does in fact compile.
   */

  @SuppressWarnings("unused") // compilation test
  public void testSampleCodeError1() {
    Optional<Integer> optionalInt = getSomeOptionalInt();
    // Number value = optionalInt.or(0.5); // error

# Separate OpenAPI Schemas for Input and Output or Not

When using **Pydantic v2**, the generated OpenAPI is a bit more exact and **correct** than before. 😎

In fact, in some cases, it will even have **two JSON Schemas** in OpenAPI for the same Pydantic model, for input and output, depending on if they have **default values**.

Let's see how that works and how to change it if you need to do that.

## Pydantic Models for Input and Output