System information

Latest main branch

What is the problem that this feature solves?

These mapping variables are implementation details that users should not be worried about them. Instead, users can simply use onnx.helper to do the conversion. See examples: https://github.com/onnx/onnx/blob/main/docs/PythonAPIOverview.md#manipulating-tensorproto-and-numpy-array. We didn't move mapping.py into _mapping.py right away because it will possibly break something. For now, using variables will throw a deprecation warning and users are encouraged to use functions from helper.instead. After several releases, we should be able to move these variables from mapping.py into _mapping.py.

Alternatives considered

Keep the original way but usually users are confused about these implementation details.

Describe the feature

To simplify and prevent confusion , only expose necessary functions for users to use.

Will this influence the current api (Y/N)?

Y

Feature Area

utilities

Are you willing to contribute it (Y/N)

Yes

Notes

No response