advanced use cases can also write their own custom
      `tf.train.experimental.ShardingCallback`s.

* `tf.train.CheckpointOptions`
    * Added `experimental_skip_slot_variables` (a boolean option) to skip
    restoring of optimizer slot variables in a checkpoint.

*   `tf.saved_model.SaveOptions`

    * `SaveOptions` now takes a new argument called

def set_windows_build_flags(environ_cp):
  """Set Windows specific build options."""

  # First available in VS 16.4. Speeds up Windows compile times by a lot. See
  # https://groups.google.com/a/tensorflow.org/d/topic/build/SsW98Eo7l3o/discussion
  # pylint: disable=line-too-long
  write_to_bazelrc(

# END TF TEST SUITE OPTIONS

# START CROSS-COMPILE CONFIGS
# Set execution platform to Linux x86
# Note: Lot of the "host_" flags such as "host_cpu" and "host_crosstool_top"
# flags seem to be actually used to specify the execution platform details. It
# seems it is this way because these flags are old and predate the distinction

# Recommended: use a local+remote cache.
#
#   Bazel will cache your builds in tensorflow/build_output/cache,
#   and will also try using public build cache results to speed up
#   your builds. This usually saves a lot of time, especially when
#   re-running tests. However, note that:
#
#    - New environments like new CUDA versions, changes to manylinux,
#      compilers, etc. can cause undefined behavior such as build failures

if [[ -z "${TFCI:-}" ]]; then
  echo '==TFCI==: The $TFCI variable is not set. This is fine as long as you'
  echo 'already sourced a TFCI env file with "set -a; source <path>; set +a".'
  echo 'If you have not, you will see a lot of undefined variable errors.'
else
  FROM_ENV=$(mktemp)
  # "export -p" prints a list of environment values in a safe-to-source format,
  # e.g. `declare -x TFCI_BAZEL_COMMON_ARGS="list of args"` for bash.