Support None as op input in GraphBuilder

[justinchuby]

Summary

Support `None` as a valid input to `op.OpType()` calls in the onnxscript builder. This is needed for ONNX ops with optional inputs (e.g., Gemm's optional C input).

Changes

• Add `None` to the `VALUE_LIKE` union type
• Update `_input_to_ir_value` to return `ir.Value | None` and pass `None` through
• Update `_partition_inputs_attributes` and `call_op` type signatures to accept `None`
• Add tests for `None` inputs with standard and custom domain ops

Also removed pylint because it is not doing much good nowadays.

[Copilot]

Pull request overview

Adds support for passing None as an op input in onnxscript’s imperative GraphBuilder, enabling correct construction of nodes with optional inputs (e.g., Gemm’s optional C input).

Changes:

• Extends VALUE_LIKE and input conversion to allow None to pass through as an omitted/optional input.
• Updates builder call paths (_partition_inputs_attributes, call_op) to accept None in input sequences.
• Adds unit tests validating None behavior for both standard-domain and custom-domain ops.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
onnxscript/_internal/builder.py	Extends accepted input types and conversion/call signatures to preserve None inputs.
onnxscript/_internal/builder_test.py	Adds tests asserting None is preserved in node input slots for optional inputs.

Comments suppressed due to low confidence (1)

onnxscript/_internal/builder.py:366

• _partition_inputs_attributes can return inputs containing None (e.g., when schema is None it returns inputs unchanged). The return type annotation currently excludes None, which makes the typing inconsistent with the new optional-input behavior and with _cast_inputs (which returns Sequence[ir.Value | None]). Please update the return type (and ideally the accepted inputs type) to include None/VALUE_LIKE so the signatures match actual runtime behavior.

        inputs: Sequence[ir.Value | ir.TensorProtocol | None],
        kwargs: dict[str, Any],
    ) -> tuple[Sequence[ir.Value | ir.TensorProtocol], dict[str, Any]]:
        if schema is None:
            return inputs, kwargs

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 71.90%. Comparing base (847801c) to head (ca315b5).
⚠️ Report is 1 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2868      +/-   ##
==========================================
+ Coverage   71.87%   71.90%   +0.02%     
==========================================
  Files         239      239              
  Lines       29150    29175      +25     
  Branches     2875     2876       +1     
==========================================
+ Hits        20953    20978      +25     
  Misses       7219     7219              
  Partials      978      978              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.