Files
tensorflow/third_party/py/python_init_toolchains.bzl
Yurii Topin 9f6a3bf72c Use Hermetic Python from rules_ml_toolchain
All essential C++, Python, CUDA, and SYCL (oneAPI) toolchains have been consolidated into rules_ml_toolchain. This repository now includes dedicated tests to ensure early bug detection whenever toolchain changes occur.

To ensure backward compatibility, logic in specific @xla//third_party/py files has been replaced with stubs. This allows dependent projects to transition to the new files gradually without breaking. Additionally, the initialization of rules_python and related repositories (including necessary patches) will remain within XLA. This is because MODULE.bazel initialization must rely strictly on local repository files and cannot utilize remote resources.

PiperOrigin-RevId: 901027238
2026-04-16 19:51:01 -07:00

18 lines
459 B
Python

"""
Legacy-compatible wrapper for python toolchain registration.
See the WORKSPACE file for instructions on using the updated
python toolchain registration.
"""
load(
"@rules_ml_toolchain//py:python_register_toolchain.bzl",
"python_register_toolchain",
)
def python_init_toolchains(name = "python", python_version = None, **kwargs):
python_register_toolchain(
name = name,
python_version = python_version,
**kwargs
)