---

**Note**: You can mix `def` and `async def` in your *path operation functions* as much as you need and define each one using the best option for you. FastAPI will do the right thing with them.

Anyway, in any of the cases above, FastAPI will still work asynchronously and be extremely fast.

But by following the steps above, it will be able to do some performance optimizations.

## Technical Details { #technical-details }

                return converter.getAsObject((String) value);
            }
            return converter.getAsString(value);
        } catch (final Throwable cause) {
            throw new ConverterRuntimeException(destPropertyName, value, cause);
        }
    }

    /**
     * Finds the converter corresponding to the class.
     *
     * @param clazz
     *            The class

<div class="screenshot">
<img src="/img/tutorial/header-param-models/image01.png">
</div>

## Forbid Extra Headers { #forbid-extra-headers }

In some special use cases (probably not very common), you might want to **restrict** the headers that you want to receive.

You can use Pydantic's model configuration to `forbid` any `extra` fields:

            .removalListener(removalListener)
            .build(loader);

    // caches 2
    assertThat(cache.getUnchecked(2)).isEqualTo(2);
    assertThat(cache.asMap().keySet()).containsExactly(2);

    CacheTesting.processPendingNotifications(cache);
    assertThat(removalListener.getCount()).isEqualTo(0);

    // caches 3, evicts 2
    assertThat(cache.getUnchecked(3)).isEqualTo(3);

import com.google.common.annotations.J2ktIncompatible;
import com.google.common.base.Function;
import org.jspecify.annotations.NullMarked;
import org.jspecify.annotations.Nullable;

/**
 * Test cases for {@link Tables#transformValues}.
 *
 * @author Jared Levy
 */
@GwtCompatible
@NullMarked
public class TablesTransformValuesTest extends AbstractTableTest<Character> {

   * to this supplier.
   *
   * <p><b>Warning: do not depend</b> on the behavior of this method.
   *
   * <p>Historically, {@code Supplier} instances in this library have implemented this method to
   * recognize certain cases where distinct {@code Supplier} instances would in fact behave
   * identically. However, as code migrates to {@code java.util.function}, that behavior will
   * disappear. It is best not to depend on it.
   */
  @Override

    void testMalformedToken() {
        byte[] bad = new byte[] { 0x01, 0x02, 0x03 }; // not a valid ASN.1 tagged object
        assertThrows(IOException.class, () -> new NegTokenTarg(bad), "Malformed byte[] should cause IOException");
    }

    @Test
    @DisplayName("mechanism token empty array – preserved after round‑trip")
    void testEmptyMechanismToken() throws IOException {
        NegTokenTarg nt = new NegTokenTarg();

        }
      }
    } finally {
      cancelInFlightConnects()
    }

    throw firstException!!
  }

  /**
   * Returns non-null if we don't need to wait for the launched result. In such cases, this result
   * must be processed before whatever is waiting in the queue because we may have already acquired
   * its connection.
   */
  private fun launchTcpConnect(): ConnectResult? {
    val plan =

    /**
     * Resolves the transitive dependencies of the specified projects. Note that dependencies which can't be resolved
     * from any repository but are present among the set of specified projects will not cause an exception. Instead,
     * those unresolved artifacts will be returned in the result set, allowing the caller to take special care of
     * artifacts that haven't been build yet.
     *

/// tip

Here we use <a href="https://docs.python.org/3/library/time.html#time.perf_counter" class="external-link" target="_blank">`time.perf_counter()`</a> instead of `time.time()` because it can be more precise for these use cases. 🤓

///

## Multiple middleware execution order { #multiple-middleware-execution-order }