// can be disabled via the corresponding `s3HFlag`.
//
// CAUTION: for requests involving large req/resp bodies ensure to pass the
// `traceHdrsS3HFlag`, otherwise both headers and body will be traced, causing
// high memory usage!
func s3APIMiddleware(f http.HandlerFunc, flags ...s3HFlag) http.HandlerFunc {
	// Collect all flags with bitwise-OR and assign operator
	var handlerFlags s3HFlag
	for _, flag := range flags {

The `string()` method on response body is convenient and efficient for small documents. But if the response body is large (greater than 1 MiB), avoid `string()` because it will load the entire document into memory. In that case, prefer to process the body as a stream.

=== ":material-language-kotlin: Kotlin"
    ```kotlin
      private val client = OkHttpClient()

      fun run() {
        val request = Request.Builder()

   * list (counter-intuitive, but mathematically consistent).
   *
   * <p><i>Performance notes:</i> while the cartesian product of lists of size {@code m, n, p} is a
   * list of size {@code m x n x p}, its actual memory consumption is much smaller. When the
   * cartesian product is constructed, the input lists are merely copied. Only as the resulting list
   * is iterated are the individual lists created, and these are not retained after iteration.

* Get windows kernel version directly from registry ([#58498](https://github.com/kubernetes/kubernetes/pull/58498), [@feiskyer](https://github.com/feiskyer))
* Fixes a bug where kubelet crashes trying to free memory under memory pressure ([#58574](https://github.com/kubernetes/kubernetes/pull/58574), [@yastij](https://github.com/yastij))

* Check if container memory stats are available before accessing it ([#77656](https://github.com/kubernetes/kubernetes/pull/77656), [@yastij](https://github.com/yastij))

   * hashmaps. Using a static ThreadLocal to avoid that overhead would mean that different
   * ExecutionSequencer objects interfere with each other, which would be undesirable, in addition
   * to increasing the memory footprint of every thread that interacted with it. In order to release
   * entries in thread-specific maps when the ThreadLocal object itself is no longer referenced,

            "maven.versionRangeResolver.natureOverride";

    /**
     * Comma-separated list of XML contexts/fields to intern during POM parsing for memory optimization.
     * When not specified, a default set of commonly repeated contexts will be used.
     * Example: "groupId,artifactId,version,scope,type"
     *
     * @since 4.0.0
     */
    @Config

			memlimit = ctx.GlobalString("memlimit")
		}
		mlimit, err := humanize.ParseBytes(memlimit)
		if err != nil {
			return err
		}
		if mlimit > memAvailable {
			logger.Info("WARNING: maximum memory available (%s) smaller than specified --memlimit=%s, ignoring --memlimit value",
				humanize.IBytes(memAvailable), memlimit)
		}
		ctxt.MemLimit = mlimit
	} else {
		ctxt.MemLimit = memAvailable
	}

- InPlacePodVerticalScaling: Memory limits cannot be decreased unless the memory resize restart policy is set to `RestartContainer`. Container resizePolicy is no longer mutable. ([#130183](https://github.com/kubernetes/kubernetes/pull/130183), [@tallclair](https://github.com/tallclair)) [SIG Apps and Node]

        }

        return currentCount <= maxEventsPerWindow;
    }

    private void cleanupOldRateLimitCounters(long currentWindow) {
        // Clean up counters from previous windows to prevent memory leaks
        rateLimitCounters.entrySet().removeIf(entry -> {
            String key = entry.getKey();
            int lastUnderscore = key.lastIndexOf('_');
            if (lastUnderscore > 0) {
                try {