# This Workflow updates tensorflow/tools/toolchains/remote_config/configs.bzl
# to reference the most recent versions of the SIG Build Docker images.
name: Update RBE Configs
on:
  workflow_dispatch:

permissions:
  contents: read

jobs:
  rbe:
    name: Update RBE Configs
    runs-on: ubuntu-latest
    if: github.repository == 'tensorflow/tensorflow' # Don't do this in forks
    steps:

    # Windows.
    environ_cp['TF_DOWNLOAD_CLANG'] = '0'
    environ_cp['TF_NEED_MPI'] = '0'

  if is_macos():
    environ_cp['TF_NEED_TENSORRT'] = '0'

  if is_ppc64le():
    # Enable MMA Dynamic Dispatch support if 'gcc' and if linker >= 2.35
    gcc_env = get_gcc_compiler(environ_cp)
    if gcc_env is not None:

      # Use gold linker if 'gcc' and if 'ppc64le'

TF_CAPI_EXPORT extern void TFE_DeleteExecutor(TFE_Executor*);

// Returns true if the executor is in async mode.
TF_CAPI_EXPORT extern bool TFE_ExecutorIsAsync(TFE_Executor*);

// Causes the calling thread to block till all ops dispatched in this executor
// have been executed. Note that "execution" here refers to kernel execution /
// scheduling of copies, etc. Similar to sync execution, it doesn't guarantee

    register a dataset with the tf.data service, and another process to consume
    data from the dataset.
*   Adds support for dispatcher fault tolerance. To enable fault tolerance,
    configure a `work_dir` when running your dispatcher server and set
    `dispatcher_fault_tolerance=True`. The dispatcher will store its state to
    `work_dir`, so that on restart it can continue from its previous state after
    restart.

        "//tensorflow/core/common_runtime/eager:tensor_handle",
        "//tensorflow/core/config:flag_defs",
        "//tensorflow/core/config:flags",
        "//tensorflow/core/data/service:dispatcher_client",
        "//tensorflow/core/data/service:grpc_util",
        "//tensorflow/core/data/service:py_utils",
        "//tensorflow/core/data/service:server_lib",
        "//tensorflow/core/debug",