// chosen for the operation by the device placement logic in the
  // executor. After that, the value returned by DeviceName will be a full
  // device name such as "/job:localhost/replica:0/task:0/device:GPU:1".
  virtual const string& DeviceName() const = 0;

  // Sets the operation device name.
  //
  // The given `name` must be parseable by DeviceNameUtils::ParseFullName, and

# /---------------------------------------------------------------------------
# basicInfoMap: (Required)
#
# The basic information for the tasks of DBFlute.
# You should specify before your first generating.
#
# Core Properties:
# o database: (Required)
# o targetLanguage: (Required)
# o targetContainer: (Required)
# o packageBase: (Required)
#
# Adjustment Properties:

There's a lot of work to do, and for most of it, **YOU** can do it.

The main tasks that you can do right now are:

* [Help others with questions in GitHub](#help-others-with-questions-in-github){.internal-link target=_blank} (see the section above).
* [Review Pull Requests](#review-pull-requests){.internal-link target=_blank} (see the section above).

Those two tasks are what **consume time the most**. That's the main work of maintaining FastAPI.

Starlette et **FastAPI** sont basés sur
<a href="https://anyio.readthedocs.io/en/stable/" class="external-link" target="_blank">AnyIO</a>, qui les rend
compatibles avec <a href="https://docs.python.org/3/library/asyncio-task.html" class="external-link" target="_blank">asyncio</a>, de la bibliothèque standard Python et
<a href="https://trio.readthedocs.io/en/stable/" class="external-link" target="_blank">Trio</a>.

    TFE_Context* ctx, const char* task, int64_t timeout_in_ms,
    TF_Status* status) {
  tensorflow::EagerContext* context =
      tensorflow::ContextFromInterface(tensorflow::unwrap(ctx));
  auto collective_executor_handle = context->GetCollectiveExecutorHandle();
  tensorflow::Notification done;
  collective_executor_handle->get()->remote_access()->CheckPeerHealth(
      task, timeout_in_ms, [&done, status](const Status& s) {

    job_def->mutable_tasks()->insert(
        {i, tensorflow::strings::StrCat("localhost:", port)});
    LOG(INFO) << "Picked test port: " << port << " for job: " << job_name
              << ", task: " << i;
  }
  return server_def;
}

tensorflow::ServerDef GetServerDef(int num_tasks) {
  return GetServerDef("localhost", num_tasks);
}

///

## Hypercorn with Trio

Starlette 和 **FastAPI** 基于 <a href="https://anyio.readthedocs.io/en/stable/" class="external-link" target="_blank">AnyIO</a>， 所以它们才能同时与 Python 的标准库 <a href="https://docs.python.org/3/library/asyncio-task.html" class="external-link" target="_blank">asyncio</a> 和<a href="https://trio.readthedocs.io/en/stable/" class="external-link" target="_blank">Trio</a> 兼容。

topologies mix and match these two.

For a standard Kubernetes deployment, both CA and discovery will use JWT authentication, with a token automatically
[generated and rotated by Kubernetes](https://kubernetes.io/docs/tasks/configure-pod-container/configure-service-account/#service-account-token-volume-projection).

For discovery, the JWT token will be read directly from a file and sent as is. For CA, this logic is a bit more complex,

    public void destroy() {
        shutdownHookList.forEach(action -> {
            try {
                action.run();
            } catch (final Exception e) {
                logger.warn("Failed to process shutdown task.", e);
            }
        });
    }

    public String getUsername() {
        final RequestManager requestManager = ComponentUtil.getRequestManager();

* Le support des **WebSockets**.
* Le support de **GraphQL**.
* Les <abbr title="En anglais: In-process background tasks">tâches d'arrière-plan.</abbr>
* Des évènements de démarrages et d'arrêt.
* Un client de test basé sur `request`
* **CORS**, GZip, Static Files, Streaming responses.
* Le support des **Sessions et Cookies**.