*   <li>Filters dependencies to include only those with transitive scopes (compile/runtime)</li>
 *   <li>Applies managed dependency metadata (version, scope, optional flag, exclusions) to direct dependencies</li>
 *   <li>Removes managed dependencies that are not used by direct dependencies</li>
 *   <li>Retains only managed dependencies that appear in the resolved dependency tree</li>
 * </ul>
 * <p>

        }
    }

    /**
     * Allocate a direct buffer for temporary use
     *
     * @param size buffer size in bytes
     * @return direct byte buffer
     */
    public ByteBuffer allocateBuffer(int size) {
        return ByteBuffer.allocateDirect(size);
    }

    /**
     * Release a temporary buffer
     *
     * For direct buffers, we rely on GC. A more sophisticated

* **Uvicorn**:
    * Wird die beste Performanz haben, da außer dem Server selbst nicht viel zusätzlicher Code vorhanden ist.

 */
package jcifs.internal.smb2.rdma;

/**
 * RDMA channel information for SMB Direct operations.
 *
 * This class encapsulates RDMA channel metadata including remote memory keys,
 * addresses, and lengths for direct memory access operations.
 */
public class RdmaChannelInfo {

    private final Smb2RdmaTransform transform;

    /**

 * <p>The main purpose of {@link #addListener addListener} is to support this chaining. You will
 * rarely use it directly, in part because it does not provide direct access to the {@code Future}
 * result. (If you want such access, you may prefer {@link Futures#addCallback
 * Futures.addCallback}.) Still, direct {@code addListener} calls are occasionally useful:
 *
 * {@snippet :
 * final String name = ...;
 * inFlight.add(name);

        include:
          - os: macos-latest
            python-version: "3.10"
            coverage: coverage
            uv-resolution: lowest-direct
          - os: windows-latest
            python-version: "3.12"
            coverage: coverage
            uv-resolution: lowest-direct
          - os: ubuntu-latest
            python-version: "3.13"
            coverage: coverage
            uv-resolution: highest

Nesse caso, você pode simplesmente trocar as `dataclasses` padrão por `pydantic.dataclasses`, que é um substituto direto:

{* ../../docs_src/dataclasses_/tutorial003_py310.py hl[1,4,7:10,13:16,22:24,27] *}

1. Ainda importamos `field` das `dataclasses` padrão.

2. `pydantic.dataclasses` é um substituto direto para `dataclasses`.

3. A dataclass `Author` inclui uma lista de dataclasses `Item`.

    app = FastAPI()

    @app.get("/")
    def endpoint(response: Response):
        response.headers["X-Direct"] = "set"
        return {"status": "ok"}

    client = TestClient(app)
    resp = client.get("/")

    assert resp.status_code == 200
    assert resp.json() == {"status": "ok"}
    assert resp.headers.get("X-Direct") == "set"


def test_response_dependency_chain():

 * <p>The main purpose of {@link #addListener addListener} is to support this chaining. You will
 * rarely use it directly, in part because it does not provide direct access to the {@code Future}
 * result. (If you want such access, you may prefer {@link Futures#addCallback
 * Futures.addCallback}.) Still, direct {@code addListener} calls are occasionally useful:
 *
 * {@snippet :
 * final String name = ...;
 * inFlight.add(name);

     * indirectly depends on the given project.
     *
     * @param project The project whose downstream projects should be retrieved, must not be {@code null}.
     * @param transitive A flag whether to retrieve all direct and indirect downstream projects or just the immediate
     *            downstream projects.
     * @return The downstream projects in the build order, never {@code null}.
     */