aglinxinyuan commented on PR #6184: URL: https://github.com/apache/texera/pull/6184#issuecomment-4888210578
Tested the cap to `1.7.2`. **CI** — the previously-failing `pyamber` legs are all green now: | leg | before | after | |---|---|---| | pyamber 3.10 | ❌ couldn't resolve `1.9.0` (needs py>=3.11) | ✅ pass | | pyamber 3.11 | ❌ cancelled (fail-fast) | ✅ pass | | pyamber 3.12 (drives the binary-license check) | ❌ cancelled | ✅ pass | | pyamber 3.13 | ❌ cancelled | ✅ pass | The 3.12 leg passing confirms the `LICENSE-binary-python` sync — `check_binary_deps.py` sees no direct-dep drift and no added/removed packages. **Local** — fresh venv on Python 3.12, `pip install scikit-learn==1.7.2`: - Dependency closure is exactly `joblib, numpy, scipy, threadpoolctl` — same four already claimed in `LICENSE-binary-python`, no new transitive dep (`narwhals` only lands in 1.9.0), so nothing trips the license check. - Instantiated all 25 classifier + 3 regressor estimators the `sklearn` operators emit, with the bare default ctor the template uses (`<Model>()`). - `make_pipeline(...).fit/predict` works, including `AdaBoostClassifier()` — the one real API risk, since 1.6 removed `algorithm='SAMME.R'`; the operators rely on the default so they're fine. - Text pipeline (`CountVectorizer -> TfidfTransformer -> classifier`) and the regression metrics (`root_mean_squared_error`, `mean_absolute_error`, `r2_score`) all present and working. `amber-integration` + `Bench` are still running — both unrelated to this bump and green before it. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
