/**
   * Alternating, decoded query names and values, or null for no query. Names may be empty or
   * non-empty, but never null. Values are null if the name has no corresponding '=' separator, or
   * empty, or non-empty.
   */
  private val queryNamesAndValues: List<String?>?,
  /**
   * This URL's fragment, like `"abc"` for `http://host/#abc`. This is null if the URL has no
   * fragment.
   *

          Object[] sortedValues = new Object[size];

          // We might, somehow, be able to reorder values in-place.  But it doesn't seem like
          // there's a way around creating the separate sortedKeys array, and if we're allocating
          // one array of size n, we might as well allocate two -- to say nothing of the allocation
          // done in Arrays.sort.
          for (int i = 0; i < size; i++) {

  released as a Pip package from TF 2.17 onwards.

### Known Caveats

*  Full aarch64 Linux and Arm64 macOS wheels are now published to the
  `tensorflow` pypi repository and no longer redirect to a separate package.

### Major Features and Improvements

*  Support for Python 3.12 has been added.
*  [tensorflow-tpu](https://pypi.org/project/tensorflow-tpu/) package is now
   available for easier TPU based installs.

* Add `--enable-admission-plugins` `--disable-admission-plugins` flags and deprecate `--admission-control`. When using the separate flag, the order in which they're specified doesn't matter.  ([#58123](https://github.com/kubernetes/kubernetes/pull/58123), [@hzxuzhonghu](https://github.com/hzxuzhonghu))

    * New docs [Separate OpenAPI Schemas for Input and Output or Not](https://fastapi.tiangolo.com/how-to/separate-openapi-schemas/).
    * This PR also includes a new setup (internal tools) for generating screenshots for the docs.

### Refactors

- The DRA device taints and toleration feature received a separate feature gate, `DRADeviceTaintRules`, which controlled support for `DeviceTaintRules`. This allowed disabling it while keeping `DRADeviceTaints` enabled so that tainting via `ResourceSlices` continued to work. ([#135068](https://github.com/kubern...

    and share it throughout the application. Requests that needs a customized
    client should call `OkHttpClient.newBuilder()` on that shared instance.
    This allows customization without the drawbacks of separate connection
    pools.

 *  **OkHttpClient is now stateless.** In the 2.x API `OkHttpClient` had getters
    and setters. Internally each request was forced to make its own complete

* Use separate client for leader election in scheduler to avoid starving leader election by regular scheduler operations. ([#53793](https://github.com/kubernetes/kubernetes/pull/53793), [@wojtek-t](https://github.com/wojtek-t))

- Fluentd: isolate logging resources in separate namespace `logging` ([#68004](https://github.com/kubernetes/kubernetes/pull/68004), [@saravanan30erd](https://github.com/saravanan30erd))

    * ServiceDelimiter string Delimiter to use to separate overridden services (multiple services) Defaults to "&"
    * ServicenameDelimiter string Delimiter to use to separate servicename from its configuration parameters Defaults "|"
    * OverrideSeparator string Delimiter to use to separate region of occurrence, url and signing region for each override Defaults to ","