for each. This changes `Address` to no longer use `ConnectionSpec`. (This is
    an incompatible API change).

 *  **`FormEncodingBuilder` now uses `%20` instead of `+` for encoded spaces.**
    Both are permitted-by-spec, but `%20` requires fewer special cases.

 *  **Okio has been updated to 1.4.0.**
     ```xml
     <dependency>
       <groupId>com.squareup.okio</groupId>
       <artifactId>okio</artifactId>

## Coroutines

**Coroutine** is just the very fancy term for the thing returned by an `async def` function. Python knows that it is something like a function, that it can start and that it will end at some point, but that it might be paused ⏸ internally too, whenever there is an `await` inside of it.

   ([#122729](https://github.com/kubernetes/kubernetes/pull/122729), [@jparrill](https://github.com/jparrill))

Et c'est le même niveau de performance que celui obtenu avec **FastAPI**.

   *       simply switch over to use the JDK implementation wholesale if probable hash flooding is
   *       detected, sacrificing the compactness guarantee in very rare cases in exchange for much
   *       more reliable worst-case behavior.
   *   <li>null, if no entries have yet been added to the map
   * </ul>
   */
  @CheckForNull private transient Object table;

  /**

 *       allocating garbage).
 * </ul>
 *
 * <p>Disadvantages compared to {@code long[]}:
 *
 * <ul>
 *   <li>Memory footprint has a fixed overhead (about 24 bytes per instance).
 *   <li><i>Some</i> construction use cases force the data to be copied (though several construction
 *       APIs are offered that don't).
 *   <li>Can't be passed directly to methods that expect {@code long[]} (though the most common

         * work with all JDK versions, TypeResolver must return the appropriate TypeVariable
         * implementation in each of the three possible cases:
         *
         * 1. Prior to JDK 7u51, the JDK TypeVariable implementation interoperates with ours.
         * Therefore, we can always create our own TypeVariable.
         *

 *       allocating garbage).
 * </ul>
 *
 * <p>Disadvantages compared to {@code int[]}:
 *
 * <ul>
 *   <li>Memory footprint has a fixed overhead (about 24 bytes per instance).
 *   <li><i>Some</i> construction use cases force the data to be copied (though several construction
 *       APIs are offered that don't).
 *   <li>Can't be passed directly to methods that expect {@code int[]} (though the most common

  // threadpool, so that we avoid the possibility of running the runner_ in the
  // threadpool of GPU event mgr, as that can trigger more callbacks to be
  // scheduled on that same threadpool, causing a deadlock in cases where the
  // caller of event_mgr->ThenExecute() blocks on the completion of the callback
  // (as in the case of ConstOp kernel creation on GPU, which involves copying a
  // CPU tensor to GPU).

### Cloud Provider

- Fixed a bug that caused PV allocation on non-English vSphere installations to fail  ([#73115](https://github.com/kubernetes/kubernetes/pull/73115), [@alvaroaleman](https://github.com/alvaroaleman))

### Cluster Lifecycle