for {
		select {
		case <-periodicTimer.C:
			now := UTCNow()
			ahs.Lock()
			for path, h := range ahs.healSeqMap {
				if h.hasEnded() && h.endTime.Add(keepHealSeqStateDuration).Before(now) {
					delete(ahs.healSeqMap, path)
				}
			}
			ahs.Unlock()

			periodicTimer.Reset(time.Minute * 5)
		case <-ctx.Done():
			// server could be restarting - need
			// to exit immediately

   * example during quiescent points between multithreaded computations. If there are updates
   * concurrent with this method, the returned value is not guaranteed to be the final
   * value occurring before the reset.
   *
   * @return the sum
   */
  public long sumThenReset() {
    long sum = base;
    Cell[] as = cells;
    base = 0L;
    if (as != null) {
      int n = as.length;

   * of tasks submitted by {@code invokeAll} or {@code invokeAny}, tasks will run serially on the
   * calling thread. Tasks are run to completion before a {@code Future} is returned to the caller
   * (unless the executor has been shutdown).
   *
   * <p>The returned executor is backed by the executor returned by {@link

///

/// note | 技術細節

你也可以使用 `from starlette.requests import Request`。

**FastAPI** 為了方便開發者而提供了它，但實際上它直接來自 Starlette。

///

### 在 `response` 之前與之後 { #before-and-after-the-response }

你可以在任何*路徑操作*接收 `request` 之前，加入要執行的程式碼。

也可以在產生出 `response` 之後、回傳之前執行程式碼。

例如，你可以新增一個自訂標頭 `X-Process-Time`，其內容為處理請求並產生回應所花費的秒數：

                jarTask.doFirst(new Action<Task>() {
                    @Override
                    public void execute(Task task) {
                        // this doFirst is added before the info plugin, therefore it will run
                        // after the doFirst added by the info plugin, and we can override attributes
                        jarTask.getManifest()
                            .attributes(

  int num_dims = TF_NumDims(t);
  for (int i = 0; i < num_dims; i++) {
    out_dims[i] = TF_Dim(t, i);
  }
}

// Runs model as is if output is a scalar,
// else sums the output tensor before returning.
absl::Status RunAndMaybeSum(AbstractContext* ctx, Model forward,
                            absl::Span<AbstractTensorHandle* const> inputs,
                            absl::Span<AbstractTensorHandle*> outputs,

{* ../../docs_src/header_params/tutorial002_an_py310.py hl[10] *}

/// warning | Achtung

Bevor Sie `convert_underscores` auf `False` setzen, bedenken Sie, dass manche HTTP-Proxys und Server die Verwendung von Headern mit Unterstrichen nicht erlauben.

///

## Doppelte Header { #duplicate-headers }

#### Which issue(s) this PR is related to:
<!--
Please link relevant issues to help with tracking.

To automatically close the linked issue(s) when this PR is merged,
add the word "Fixes" before the issue number or link.
Do not use "Fixes" if the PR is of kind `failing-test` or `flake`.

Reference KEPs when applicable in addition to specific issues.

Examples:
Fixes #<issue number>

     */
    protected boolean done = false;

    /**
     * Constructs a ContentOutputStream with a specified threshold and temporary directory.
     *
     * @param threshold the threshold in bytes for writing to memory before writing to a file
     * @param tmpDir the temporary directory to create the file in
     */
    public ContentOutputStream(final int threshold, final File tmpDir) {
        super(threshold, PREFIX, SUFFIX, tmpDir);
    }

  ) {
    logWithTime("followUpDecision")
  }

  private fun logWithTime(message: String) {
    val startNs = startNs
    val timeMs =
      if (startNs == null) {
        // Event occurred before start, for an example an early cancel.
        0L
      } else {
        TimeUnit.NANOSECONDS.toMillis(System.nanoTime() - startNs)
      }

    logger.invoke("[$timeMs ms] $message")
  }