[jrmccluskey]

Experimental LLM driven update of our ML dependency extra lists to modernize the suite and get more of these tests running regularly again.

DO NOT MERGE

[codecov]

Codecov Report

❌ Patch coverage is 0% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 56.46%. Comparing base (a8d6afd) to head (feec563).
⚠️ Report is 74 commits behind head on master.

Files with missing lines	Patch %	Lines
sdks/python/setup.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@              Coverage Diff              @@
##             master   #40531       +/-   ##
=============================================
- Coverage     80.04%   56.46%   -23.59%     
- Complexity        0     2288     +2288     
=============================================
  Files           490     1127      +637     
  Lines         82916   178924    +96008     
  Branches          0     1489     +1489     
=============================================
+ Hits          66373   101025    +34652     
- Misses        16543    75378    +58835     
- Partials          0     2521     +2521     

Flag	Coverage Δ
python	80.02% <0.00%> (-0.03%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[jrmccluskey]

/gemini review

[github-actions]

📋 Review Summary

This PR updates the Python ML test dependency extras in sdks/python/setup.py to re-enable tensorflow_transform (1.21.x), xgboost (>=2.1.3 on Python 3.11–3.12), and google-adk (2.x) alongside Python 3.14 environment markers, while fixing two downstream test compatibility issues with newer XGBoost and NumPy versions. The dependency bounds and test fixes are well-reasoned, and there are a couple of minor cleanup items in setup.py to keep the extra definitions consistent.

🔍 General Feedback

• The test fixes in xgboost_inference_test.py and base_test.py cleanly address attribute persistence across model reloads and NumPy >= 2.4 scalar conversion rules.
• In sdks/python/setup.py, a few related standalone/legacy extras (xgboost, ml_test, p310_ml_test) and duplicate tokenizers entries in ml_base_core could be aligned with the updated dependency bounds.

---

💡 Tip: Comment /gemini review on this pull request to trigger a new Gemini code review.

💬 Inline Findings

• sdks/python/setup.py (line 211): 🟢 Low: ml_base_core currently specifies tokenizers twice: 'tokenizers<0.23' here on line 211 and 'tokenizers>=0.13.3,<0.22.0' on line 217 (which is strictly narrower and makes <0.23 redundant). Consider consolidating these into a single tokenizers requirement entry so the version cap and accompanying comment stay in sync.

• sdks/python/setup.py (line 623): 🟢 Low: Since python_requires = '>=3.11' (and tox.ini only defines py311-ml through py314-ml), p310_ml_test appears to be unused. In addition, the standalone 'xgboost' extra further down (line 699: 'xgboost': ['xgboost>=1.6.0,<2.1.3', 'datatable==1.0.0']) still caps xgboost<2.1.3, which contradicts 'xgboost>=2.1.3' added here for p311_ml_test and p312_ml_test. Consider removing p310_ml_test and updating the 'xgboost' extra bound to match.