potentially exploit a vulnerability in the TensorFlow code that handles videos,
which could allow them to execute arbitrary code on the system running
TensorFlow.

It is important to keep TensorFlow up to date with the latest security patches
and follow the sandboxing guideline above to protect against these types of
vulnerabilities.

## Security properties of execution modes

/// the 3 types of files. In all other cases, there is no need to modify the
/// address of the opaque data pointer, hence the wrapper pointer is marked
/// `const`.
///
/// For consistency, the arguments on all these functions follow the same
/// pattern: first we have the opaque pointer argument ("this" above), then the
/// input arguments, then the in-out arguments (if any) and we finish the

    [FTRL paper](https://research.google.com/pubs/archive/41159.pdf).
*   `Optimizer.__init__` now accepts a `gradient_aggregator` to allow for
    customization of how gradients are aggregated across devices, as well as
    `gradients_transformers` to allow for custom gradient transformations (such
    as gradient clipping).
*   Improvements to Keras preprocessing layers:

// `enable_xla_compilation` is non-zero, and OFF otherwise.
// b) ConfigProto.gpu_options.allow_growth is set to `gpu_memory_allow_growth`.
// c) ConfigProto.device_count is set to `num_cpu_devices`.
TF_CAPI_EXPORT extern TF_Buffer* TF_CreateConfig(
    unsigned char enable_xla_compilation, unsigned char gpu_memory_allow_growth,
    unsigned int num_cpu_devices);

  } else {
    optimizer_options->set_global_jit_level(tensorflow::OptimizerOptions::OFF);
  }

  auto* gpu_options = config.mutable_gpu_options();
  gpu_options->set_allow_growth(gpu_memory_allow_growth);

  (*config.mutable_device_count())["CPU"] = num_cpu_devices;

  // TODO(b/113217601): This is needed for EagerContext::runner_ to use a

def write_action_env_to_bazelrc(var_name, var):
  write_to_bazelrc('build --action_env {}="{}"'.format(var_name, str(var)))


def run_shell(cmd, allow_non_zero=False, stderr=None):
  if stderr is None:
    stderr = sys.stdout
  if allow_non_zero:
    try:
      output = subprocess.check_output(cmd, stderr=stderr)
    except subprocess.CalledProcessError as e:
      output = e.output
  else:

      TFE_NewContextOptions(), TFE_DeleteContextOptions);
  std::unique_ptr<TF_Buffer, decltype(&TF_DeleteBuffer)> config(
      TF_CreateConfig(
          /*xla*/ false,
          /* gpu_memory_allow_growth */ true, /* num_cpu_devices */
          2),
      TF_DeleteBuffer);
  TFE_ContextOptionsSetConfig(opts.get(), config->data, config->length,
                              status.get());

//
// Gradient is the type returned by gradient functions. In Python TF it's either
// Tensor or IndexedSlices or None, which here we map to nullptr. Gradients need
// to allow their size to be computed and they need to be passable to a backward
// function and deleted (as the backprop code creates lots of gradients the user
// is not interested in).
//

# opts in to modular op registration support by default.
build --define framework_shared_object=true
build --define tsl_protobuf_header_only=true

build --define=use_fast_cpp_protos=true
build --define=allow_oversize_protos=true

build --spawn_strategy=standalone
build -c opt

# Make Bazel print out all options from rc files.
build --announce_rc

# TODO(mihaimaruseac): Document this option or remove if no longer needed

#### *Try your first TensorFlow program*

```shell
$ python
```

```python
>>> import tensorflow as tf
>>> tf.add(1, 2).numpy()
3
>>> hello = tf.constant('Hello, TensorFlow!')
>>> hello.numpy()
b'Hello, TensorFlow!'
```

For more examples, see the
[TensorFlow tutorials](https://www.tensorflow.org/tutorials/).

## Contribution guidelines