Skip to content

[Draft] ONNX Tensorflow Release Protocol and Checklist

Jump to bottom
Tian Jin edited this page Mar 5, 2019 · 2 revisions

This protocol is triggered every time a new ONNX minor release is available.

  • Test both locally and on Travis that the current onnx/tensorflow release version works with onnx-tensorflow (CI must be green).
  • Tag the master as a release named "onnx-[onnx release version]" (e.g., onnx-v1.3).
  • Upload the tagged master to pypi.
  • Document onnx and tf versions for onnx-tf releases in README.
  • Containerize the particular tests with ONNX release, Tensorflow release and ONNX-Tensorflow release we've tested using Docker.
  • Include a travis build status badge to the specific branch alongside the release note in README.

Optionally, if improvements to support prior onnx, tensorflow versions are needed, we will branch out based on the release tagged commit to commence improvement.

Clone this wiki locally