As you saw in the previous chapter about [Deployment Concepts](concepts.md), there are multiple strategies you can use.

Here I'll show you how to use **Uvicorn** with **worker processes** using the `fastapi` command or the `uvicorn` command directly.

/// info

/**
 * A map from artifact name to a set of class name prefixes that should be kept.
 * Artifacts matched by this map will be minified to only contain the specified
 * classes and the classes they depend on. The classes are not relocated, they all
 * remain in their original namespace. This reduces the final Gradle distribution
 * size and makes us more conscious of which parts of a library we really need.
 *

import java.util.LinkedList

/**
 * Split a list of elements into nearly even sublist. If an element is too large, largeElementSplitFunction will be used to split the large element into several smaller pieces;
 * if some elements are too small, they will be aggregated by smallElementAggregateFunction.
 *
 * @param list the list to split, must be ordered by size desc
 * @param toIntFunction the function used to map the element to its "size"

    /*
     * We use `fromNullable` instead of `of` because `WeakReference.get()` has a nullable return
     * type.
     *
     * In practice, we are very unlikely to see `null`: The `WeakReference` to the enum constant
     * won't be cleared as long as the enum constant is referenced somewhere, and the enum constant

 * invoke methods, they invoke methods on the {@code ForwardingQueue}.
 *
 * <p>The {@code standard} methods are not guaranteed to be thread-safe, even when all of the
 * methods that they depend on are thread-safe.
 *
 * @author Mike Bostock
 * @author Louis Wasserman
 * @since 2.0
 */
@GwtCompatible

 *   <li>Intern objects for faster equality comparisons</li>
 *   <li>Apply custom optimization strategies</li>
 *   <li>Transform or modify objects during building</li>
 * </ul>
 *
 * <p>Implementations are discovered via the Java ServiceLoader mechanism and should
 * be registered in {@code META-INF/services/org.apache.maven.api.model.ModelObjectProcessor}.</p>
 *

}

// hashOrder - hashes input key to return consistent
// hashed integer slice. Returned integer order is salted
// with an input key. This results in consistent order.
// NOTE: collisions are fine, we are not looking for uniqueness
// in the slices returned.
func hashOrder(key string, cardinality int) []int {
	if cardinality <= 0 {
		// Returns an empty int slice for cardinality < 0.
		return nil
	}

  }

  /**
   * Tests that the method {@code putEdge} will silently add the missing nodes to the graph, then
   * add the edge connecting them. We are not using the proxy methods here as we want to test {@code
   * putEdge} when the end-points are not elements of the graph.
   */
  @Test
  public void putEdge_nodesNotInGraph() {
    assume().that(graphIsMutable()).isTrue();

    graphAsMutableGraph.addNode(N1);

    validations:
      required: true
  - type: textarea
    id: context
    attributes:
      label: Context (optional)
      description: |
        How has this issue affected you? What are you trying to accomplish?
        Providing context helps us come up with a solution that is most useful in the real world
    validations:

 * }
 *
 * <p>Now {@code MY_SPLITTER.split("foo,,, bar ,")} returns just {@code ["foo", "bar"]}. Note that
 * the order in which these configuration methods are called is never significant.
 *
 * <p><b>Warning:</b> Splitter instances are immutable. Invoking a configuration method has no
 * effect on the receiving instance; you must store and use the new splitter instance it returns
 * instead.
 *
 * {@snippet :