I noticed a small dependency-floor drift.
pyproject.toml requires scipy>=1.10, with an inline comment explaining that lower_limit= support is needed for scipy.stats.multivariate_normal.cdf(...). Four documentation/docs-build surfaces still say or install an older SciPy floor:
- README requirements:
scipy >= 1.7
- Read the Docs post-install:
scipy>=1.7.0
- docs Sphinx build workflow:
scipy>=1.7.0
- docs dependency smoke workflow on Python 3.9:
scipy>=1.7.0
Would you accept a narrow consistency PR that updates those references to scipy>=1.10 only?
Proposed scope:
- update those four active lower-bound references;
- leave normal test workflows that install latest
scipy unchanged;
- avoid code changes, runtime metadata changes, new CI jobs, and dependency-policy changes.
Validation I will run:
- search guard for remaining active
scipy>=1.7 lower-bound claims;
- Python 3.9 docs dependency install smoke;
- docs snippet and doc-dependency integrity tests;
- Sphinx HTML build with warnings treated as errors.
I noticed a small dependency-floor drift.
pyproject.tomlrequiresscipy>=1.10, with an inline comment explaining thatlower_limit=support is needed forscipy.stats.multivariate_normal.cdf(...). Four documentation/docs-build surfaces still say or install an older SciPy floor:scipy >= 1.7scipy>=1.7.0scipy>=1.7.0scipy>=1.7.0Would you accept a narrow consistency PR that updates those references to
scipy>=1.10only?Proposed scope:
scipyunchanged;Validation I will run:
scipy>=1.7lower-bound claims;