char c1 = s1.charAt(i);
      char c2 = s2.charAt(i);
      if (c1 == c2) {
        continue;
      }
      int alphaIndex = getAlphaIndex(c1);
      // This was also benchmarked using '&' to avoid branching (but always evaluate the rhs),
      // however this showed no obvious improvement.
      if (alphaIndex < 26 && alphaIndex == getAlphaIndex(c2)) {
        continue;
      }
      return false;
    }

      // Both a and b are odd. Assume a > b; then gcd(a - b, b) = gcd(a, b).
      // But in gcd(a - b, b), a - b is even and b is odd, so we can divide out powers of two.

      // We bend over backwards to avoid branching, adapting a technique from
      // http://graphics.stanford.edu/~seander/bithacks.html#IntegerMinOrMax

      long delta = a - b; // can't overflow, since a and b are nonnegative

    var finished: Boolean = false,
  ) : Sink {
    /**
     * Buffer of outgoing data. This batches writes of small writes into this sink as larges frames
     * written to the outgoing connection. Batching saves the (small) framing overhead.
     */
    private val sendBuffer = Buffer()

    /** Trailers to send at the end of the stream. */
    var trailers: Headers? = null

    var closed: Boolean = false

### **Instrumentation**

#### **Audit**

*   Adjust batching audit webhook default parameters: increase queue size, batch size, and initial backoff. Add throttling to the batching audit webhook. Default rate limit is 10 QPS. ([#53417](https://github.com/kubernetes/kubernetes/pull/53417),[ @crassirostris](https://github.com/crassirostris))

* Webhook always retries connection reset error. ([#53947](https://github.com/kubernetes/kubernetes/pull/53947), [@crassirostris](https://github.com/crassirostris))
* Adjust batching audit webhook default parameters: increase queue size, batch size, and initial backoff. Add throttling to the batching audit webhook. Default rate limit is 10 QPS. ([#53417](https://github.com/kubernetes/kubernetes/pull/53417), [@crassirostris](https://github.com/crassirostris))

- EndpointSlice should not contain endpoints for terminating pods ([#89056](https://github.com/kubernetes/kubernetes/pull/89056), [@andrewsykim](https://github.com/andrewsykim)) [SIG Apps and Network]
- Evictions due to pods breaching their ephemeral storage limits are now recorded by the `kubelet_evictions` metric and can be alerted on. ([#87906](https://github.com/kubernetes/kubernetes/pull/87906), [@smarterclayton](https://github.com/smarterclayton)) [SIG Node]

* Log audit backend can now be configured to perform batching before writing events to disk. ([#60237](https://github.com/kubernetes/kubernetes/pull/60237), [@crassirostris](https://github.com/crassirostris))

- Kube-proxy: removed log warning about not using config file. ([#118115](https://github.com/kubernetes/kubernetes/pull/118115), [@TommyStarK](https://github.com/TommyStarK)) [SIG Network]
- Made Job controller batching of syncJob invocations enabled unconditionally (it was conditional on JobReadyPods feature before).

* Defaults for file audit logging backend in batch mode changed: ([#67223](https://github.com/kubernetes/kubernetes/pull/67223), [@tallclair](https://github.com/tallclair))
    * Logs are written 1 at a time (no batching)
    * Only a single writer process (lock contention)

- EndpointSlice should not contain endpoints for terminating pods ([#89056](https://github.com/kubernetes/kubernetes/pull/89056), [@andrewsykim](https://github.com/andrewsykim)) [SIG Apps and Network]
- Evictions due to pods breaching their ephemeral storage limits are now recorded by the `kubelet_evictions` metric and can be alerted on. ([#87906](https://github.com/kubernetes/kubernetes/pull/87906), [@smarterclayton](https://github.com/smarterclayton)) [SIG Node]