}
};

// An abstract operation describes an operation by its type, name, and
// attributes. It can be "executed" by the context with some input tensors.
// It is allowed to reusing the same abstract operation for multiple execution
// on a given context, with the same or different input tensors.
class TracingOperation : public AbstractOperation {
 protected:

 * number of streams, dynamically configured with `SETTINGS_MAX_CONCURRENT_STREAMS`. A connection
 * currently carrying zero streams is an idle stream. We keep it alive because reusing an existing
 * connection is typically faster than establishing a new one.
 *
 * When a single logical call requires multiple streams due to redirects or authorization

#### Per-Connection Locks

Each connection has its own lock. The connections in the pool are all in a `ConcurrentLinkedQueue`. Due to data races, iterators of this queue may return removed connections. Callers must check the connection's `noNewExchanges` property before using connections from the pool.

  static final String ERROR_ADDED_PARALLEL_EDGE = "Should not be allowed to add a parallel edge.";
  static final String ERROR_ADDED_EXISTING_EDGE =
      "Reusing an existing edge to connect different nodes succeeded";

  /** Creates and returns an instance of the graph to be tested. */
  abstract Network<Integer, String> createGraph();

  /**

   *     {@link #get(Object)}; may contain duplicates), in an increasing index order.
   */
  public Iterable<L> bulkGet(Iterable<? extends Object> keys) {
    // Initially using the list to store the keys, then reusing it to store the respective L's
    List<Object> result = newArrayList(keys);
    if (result.isEmpty()) {
      return ImmutableList.of();
    }
    int[] stripes = new int[result.size()];

#include "tensorflow/c/eager/c_api.h"

#ifdef __cplusplus
extern "C" {
#endif

// Resets `op_to_reset` with `op_or_function_name` and `raw_device_name`. This
// is for performance optimization by reusing an exiting unused op rather than
// creating a new op every time. If `raw_device_name` is `NULL` or empty, it
// does not set the device name. If it's not `NULL`, then it attempts to parse

        while ( iter.hasNext() ) {
            SmbSessionImpl ssn = iter.next();
            if ( ssn.matches(tf, targetHost, targetDomain) ) {
                if ( log.isTraceEnabled() ) {
                    log.trace("Reusing existing session " + ssn);
                }
                return ssn.acquire();
            }
            else if ( log.isTraceEnabled() ) {

 *
 * OkHttp performs best when you create a single `OkHttpClient` instance and reuse it for all of
 * your HTTP calls. This is because each client holds its own connection pool and thread pools.
 * Reusing connections and threads reduces latency and saves memory. Conversely, creating a client
 * for each request wastes resources on idle pools.
 *
 * Use `new OkHttpClient()` to create a shared instance with the default settings:
 *

	// is too large we are basically just thrashing the CPU D-cache.
	// So if the result length is larger than an empirically-found
	// limit (8KB), we stop growing the source string once the limit
	// is reached and keep reusing the same source string - that
	// should therefore be always resident in the L1 cache - until we
	// have completed the construction of the result.
	// This yields significant speedups (up to +100%) in cases where

				dst = make([]byte, 0, hdr.Length+hdr.Length>>3)
			}
			if !hdr.Fin {
				hdr.Length = -1
			}
			return readAllInto(dst[:0], &wsReader, hdr.Length)
		}
	}

	// Keep reusing the same buffer.
	var msg []byte
	for atomic.LoadUint32((*uint32)(&c.state)) == StateConnected {
		if cap(msg) > readBufferSize*4 {
			// Don't keep too much memory around.
			msg = nil
		}