[TIR][Schedule] Fix type checker to support subscripted generics in Python 3.14+

[khj809]

This PR fixes the type annotation checker in tvm.tir.schedule._type_checker to correctly handle subscripted generics (e.g., Union[str, int], List[str], Tuple[str, int]) in Python 3.14+.

Background

In Python 3.14, the internal representation of generic types has changed:

• Union[str, int] is now of type typing.Union instead of typing._GenericAlias or typing._SpecialGenericAlias
• These types now have __origin__ attribute directly on the type object
• The existing type checker failed to recognize these new representations, causing the dispatcher to fall through to "atomic" instead of correctly identifying them as "union", "list", etc.

Changes

Added a check for __origin__ attribute at the beginning of the method to handle Python 3.14's new generic type representations. This is fully backward compatible since the new __origin__ check is only applied when the attribute exists.

Tests

Added parametrized tests to verify the dispatcher correctly handles subscripted generics:

• Union[str, int] → identified as "union"
• List[str] → identified as "list"
• Dict[str, int] → identified as "dict"
• Tuple[str, int] → identified as "tuple"
• Union[List[str], Dict[str, int]] → identified as "union" with nested generics

[gemini-code-assist]

Summary of Changes

Hello @khj809, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request resolves a compatibility issue within the tvm.tir.schedule module's type checker, specifically concerning how it processes generic type annotations in Python 3.14 and newer. Due to internal changes in Python 3.14's representation of generic types, the existing type checker was failing to correctly identify types like Union[str, int] or List[str]. The changes introduce a robust and backward-compatible mechanism to ensure these types are correctly dispatched, preventing runtime errors and maintaining type safety across Python versions.

Highlights

• Python 3.14+ Compatibility: The type annotation checker in tvm.tir.schedule._type_checker has been updated to correctly handle subscripted generics in Python 3.14 and later versions.
• Generic Type Representation Fix: A new check for the __origin__ attribute was added to the _origin method, addressing changes in how Python 3.14 represents generic types like Union, List, and Dict.
• Backward Compatibility: The implemented fix is fully backward compatible, as the new __origin__ check is only applied when the attribute exists, ensuring older Python versions are unaffected.
• New Test Cases: Parametrized tests have been added to verify that the _dispatcher correctly identifies and processes various subscripted generic types, including nested generics, under the new Python 3.14+ type system.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

The pull request introduces a change to _type_checker.py to improve how the origin of generic types is determined, specifically adding a direct __origin__ attribute check for Python 3.14+ compatibility. Concurrently, a new test test_subscripted_generics was added to test_type_annotation_checker.py to verify the _dispatcher's handling of various generic type annotations and their subtypes, especially in the context of Python 3.14's internal representation changes. The review comment suggests enhancing this new test by making the assertion for nested Union types more robust, recommending a check for exact subtypes instead of just the count, which would also simplify the test's conditional logic.

[mshr-h]

LGTM.