* per call to submit() or submitAsync(), and each instance supports only one call to execute().
   *
   * <p>This class would certainly be simpler and easier to reason about if it were built with
   * ThreadLocal; however, ThreadLocal is not well optimized for the case where the ThreadLocal is
   * non-static, and is initialized/removed frequently - this causes churn in the Thread specific

 *  Fix: Support executing OkHttp on kotlin-stdlib versions as old as 1.4. The library still builds
    on up-to-date Kotlin releases (1.6.21) but no longer needs that version as a runtime dependency.
    This should make it easier to use OkHttp in Gradle plugins.

 *  Fix: Don't start the clock on response timeouts until the request body is fully transmitted.
    This is only relevant for duplex request bodies, because they are written concurrently when

      comment syntax for the file format. We also recommend that a
      file or class name and description of purpose be included on the
      same "printed page" as the copyright notice for easier
      identification within third-party archives.

   Copyright {yyyy} {name of copyright owner}

   Licensed under the Apache License, Version 2.0 (the "License");

Of course, there are some cases where there's no problem in running the previous steps multiple times, in that case, it's a lot easier to handle.

/// tip

Also, keep in mind that depending on your setup, in some cases you **might not even need any previous steps** before starting your application.

This is "synchronous" work, you are "synchronized" with the cashier/cook 👨‍🍳. You have to wait 🕙 and be there at the exact moment that the cashier/cook 👨‍🍳 finishes the burgers and gives them to you, or otherwise, someone else might take them.

<img src="/img/async/parallel-burgers/parallel-burgers-04.png" class="illustration">

   * invokeAny} which are pending serial execution, even the subset of the tasks that have not yet
   * started execution. It is unclear from the {@code ExecutorService} specification if these should
   * be included, and it's much easier to implement the interpretation that they not be. Finally, a
   * call to {@code shutdown} or {@code shutdownNow} may result in concurrent calls to {@code

        }
      }
    }
    for (Invokable<?, ?> factory : factories) {
      factory.setAccessible(true);
    }
    // Sorts methods/constructors with the least number of parameters first since it's likely easier
    // to fill dummy parameter values for them. Ties are broken by name then by the string form of
    // the parameter list.
    return BY_NUMBER_OF_PARAMETERS
        .compound(BY_METHOD_NAME)

        ":with_tpu_support_define",
        ":with_tpu_support_flag",
    ],
    visibility = ["//visibility:public"],
)

# Specifies via a config setting if this is a mobile build or not, makes
# it easier to combine settings later.
selects.config_setting_group(
    name = "mobile",
    match_any = [
        ":android",
        ":chromiumos",

#include "tensorflow/c/eager/c_api_internal.h"
#include "tensorflow/c/eager/immediate_execution_operation.h"
#include "tensorflow/c/eager/immediate_execution_tensor_handle.h"
#include "tensorflow/c/eager/tfe_context_internal.h"
#include "tensorflow/c/eager/tfe_op_internal.h"
#include "tensorflow/c/eager/tfe_tensorhandle_internal.h"
#include "tensorflow/c/tf_buffer_internal.h"
#include "tensorflow/c/tf_status.h"

and less like C and C++, where the meaning of any program with a race
is entirely undefined, and the compiler may do anything at all.
Go's approach aims to make errant programs more reliable and easier to debug,
while still insisting that races are errors and that tools can diagnose and report them.
</p>

<h2 id="model">Memory Model</h2>

<p>
The following formal definition of Go's memory model closely follows