const (
	// minBufferSize is the minimum buffer size.
	// Buffers below this is not reused.
	minBufferSize = 1 << 10

	// defaultBufferSize is the default buffer allocation size.
	defaultBufferSize = 4 << 10

	// maxBufferSize is the maximum buffer size.
	// Buffers larger than this is not reused.
	maxBufferSize = 96 << 10

	// This is the assumed size of bigger buffers and allocation size.

   *
   * By default OkHttp will attempt to retransmit request bodies when the original request fails
   * due to any of:
   *
   *  * A stale connection. The request was made on a reused connection and that reused connection
   *    has since been closed by the server.
   *  * A client timeout (HTTP 408).
   *  * A authorization challenge (HTTP 401 and 407) that is satisfied by the [Authenticator].

import static java.lang.annotation.RetentionPolicy.RUNTIME;

/**
 * Denotes that a bean should be created as a singleton instance.
 * <p>
 * Singleton-scoped beans are instantiated once and reused throughout the entire
 * Maven execution. This scope should be used for stateless services or components
 * that can be safely shared across the entire build process.
 * <p>
 * Example usage:
 * <pre>

- Don't use a custom dialer for the kubelet if is not rotating certificates, so we can reuse TCP connections and have only one between the apiserver and the kubelet.

  RecordedRequest request3 = server.takeRequest();
  assertEquals("/v1/chat/messages/3", request3.getUrl().encodedPath());

  // Shut down the server. Instances cannot be reused.
  server.close();
}
```

### Kotlin
```kotlin
fun test() {
  // Create a MockWebServer. These are lean enough that you can create a new
  // instance for every unit test.
  val server = MockWebServer()

        } else {
          exchange.noRequestBody()
          if (!exchange.connection.isMultiplexed) {
            // If the "Expect: 100-continue" expectation wasn't met, prevent the HTTP/1 connection
            // from being reused. Otherwise we're still obligated to transmit the request body to
            // leave the connection in a consistent state.
            exchange.noNewExchangesOnConnection()
          }
        }
      } else {

			// with dead end because tcp-keepalive is not fired when there is data in the socket buffer.
			//    https://blog.cloudflare.com/when-tcp-sockets-refuse-to-die/
			// This is a sensitive configuration, it is better to set it to high values, > 60 secs since it can
			// affect clients reading data with a very slow pace  (disappropriate with socket buffer sizes)

on the same machine by simply switching the value of `HERMETIC_PYTHON_VERSION`
between the runs. All the python-agnostic parts of the build cache from the
previous build will be preserved and reused for the subsequent builds.

### Specifying Python dependencies

During bazel build all TensorFlow's Python dependencies are pinned to their
specific versions. This is necessary to ensure reproducibility of the build.

The Gradle codebase has evolved over time and has a mixture of simple getter/setter methods, Provider API and things in between.  It can be hard to follow best practices because Gradle provided types are reused in unexpected ways and extended by third party plugins.

Given these constraints, we cannot always follow best practices.  This ADR proposes the way we should handle the use of Provider APIs in the gradle/gradle codebase.

      "Edge %s that was used to generate this set is no longer in the graph.";
  static final String REUSING_EDGE =
      "Edge %s already exists between the following nodes: %s, "
          + "so it cannot be reused to connect the following nodes: %s.";
  static final String MULTIPLE_EDGES_CONNECTING =
      "Cannot call edgeConnecting() when parallel edges exist between %s and %s. Consider calling "
          + "edgesConnecting() instead.";