Update Ort::Custom::GetOrtDType() to fix unreachable code warning.

[Edward Chen (edgchen1)]

Replace return after ORTX_CXX_API_THROW with static_assert on unsupported type.

Also added support for additional types and made the function constexpr.

[Copilot]

🟢 Approval recommended

No blocking issues were identified; additional mapping tests are a minor nit.

Pull request overview

Updates Ort::Custom::GetOrtDType() with constexpr type mapping and compile-time validation for unsupported types.

Changes:

• Adds MFloat16, BFloat16, and std::string_view mappings.
• Replaces unreachable fallback code with a dependent static_assert.

File summaries

File	Summary
include/custom_op/tensor_api.h	Updates tensor element-type mappings and unsupported-type handling.

Review details

Suppressed comments (1)

include/custom_op/tensor_api.h:150

• This changes the type mapping for MFloat16, BFloat16, and std::string_view, but test/static_test/test_tensor_api.cc has no assertions for Tensor::Type() or GetOrtDType(). Add coverage for these mappings so a regression does not silently restore the previous exception or report the wrong ONNX element type.

#if ORT_API_VERSION >= 16
  else if constexpr (std::is_same<TT, MFloat16>::value)
    return ONNX_TENSOR_ELEMENT_DATA_TYPE_FLOAT16;
  else if constexpr (std::is_same<TT, BFloat16>::value)
    return ONNX_TENSOR_ELEMENT_DATA_TYPE_BFLOAT16;

• Files reviewed: 1/1 changed files
• Comments generated: 0
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.