* HTTP/2 support allows all requests to the same host to share a socket.
 * Connection pooling reduces request latency (if HTTP/2 isn’t available).
 * Transparent GZIP shrinks download sizes.
 * Response caching avoids the network completely for repeat requests.

OkHttp perseveres when the network is troublesome: it will silently recover from common connection

	ObjectSizeGreaterThan int64 `xml:"ObjectSizeGreaterThan,omitempty"`
	ObjectSizeLessThan    int64 `xml:"ObjectSizeLessThan,omitempty"`

	And    And
	andSet bool

	Tag    Tag
	tagSet bool

	// Caching tags, only once
	cachedTags map[string]string
}

// MarshalXML - produces the xml representation of the Filter struct
// only one of Prefix, And and Tag should be present in the output.

build --linkopt="-lm"
build --linkopt="-Wl,--undefined-version"

# Prevent double-compilation of some TF code, ref. b/183279666 (internal)
# > TF's gen_api_init_files has a genrule to run the core TensorFlow code
# > on the host machine. If we don't have --distinct_host_configuration=false,
# > the core TensorFlow code will be built once for the host and once for the
# > target platform.

- **core-configuration**: Allows the build structure and work, such as tasks, to be specified. This includes the project model, the DSL and so on.
- **core-execution**: Runs the work efficiently. This includes scheduling, execution, caching and so on.

### Software development platform

This is a general purpose platform that builds on the core automation platform to add support for the automation of software development.

  ## GitHub 저장소 보기

  GitHub에서 FastAPI를 "watch"할 수 있습니다 (오른쪽 상단 watch 버튼을 클릭): https://github.com/fastapi/fastapi. 👀

  "Releases only" 대신 "Watching"을 선택하면 다른 사용자가 새로운 issue를 생성할 때 알림이 수신됩니다.

  그런 다음 이런 issues를 해결 할 수 있도록 도움을 줄 수 있습니다.

  ## 이슈 생성하기

  GitHub 저장소에 [새로운 이슈 생성](https://github.com/fastapi/fastapi/issues/new/choose) 을 할 수 있습니다, 예를들면 다음과 같습니다:

This is done by an HTTP server running on `/var/run/istio-cni/pluginevent.sock`.

An alternative flow is when a pod is enrolled into ambient mode after it starts up.
In this case, the CNI Agent is watching for Pod events from the API server directly and performing the same setup.
Note this is done while the Pod is running, unlike the CNI plugin flow which occurs before the Pod starts.

		secName := remoteSecretNameFromClusterName(c.clusterName)
		t.Run(fmt.Sprintf("[%v] %v", i, c.name), func(tt *testing.T) {
			// no updateTokenSecret means re-fetching yields the same result
			obj := []runtime.Object{c.haveTokenSecret}
			if c.updatedTokenSecret != nil {
				// fetching should give a different result than the token secret we pass in
				obj = []runtime.Object{c.updatedTokenSecret}
			}
			client := kube.NewFakeClient(obj...)

Linux-Container werden mit demselben Linux-Kernel des Hosts (Maschine, virtuellen Maschine, Cloud-Servers, usw.) ausgeführt. Das bedeutet einfach, dass sie sehr leichtgewichtig sind (im Vergleich zu vollständigen virtuellen Maschinen, die ein gesamtes Betriebssystem emulieren).

Auf diese Weise verbrauchen Container **wenig Ressourcen**, eine Menge vergleichbar mit der direkten Ausführung der Prozesse (eine virtuelle Maschine würde viel mehr verbrauchen).

     */
    open fun <T> tag(
      type: Class<in T>,
      tag: T?,
    ) = commonTag(type.kotlin, tag)

    /**
     * Override the [Request.url] for caching, if it is either polluted with
     * transient query params, or has a canonical URL possibly for a CDN.
     *
     * Note that POST requests will not be sent to the server if this URL is set

  //
  // A consequence of these requirements is that the delegate futures cannot be stored in
  // final fields.
  //
  // For simplicity the rest of this description will discuss Futures.catching since it is the
  // simplest instance, though very similar descriptions apply to many other classes in this file.
  //
  // In the constructor of AbstractCatchingFuture, the delegate future is assigned to a field