v1.0.0

What's Changed

• release: v1.0.0 by @fdosani in #516

Full Changelog: v0.19.5...v1.0.0

Summary

This is the v1.0.0 general availability release of DataComPy, a major version bump from 0.19.x. It introduces a new comparator architecture, sensitive column handling, and several quality-of-life improvements across all backends.

Breaking Changes

• Comparator subpackage (datacompy/comparator/): New strategy-pattern architecture for column comparison logic. Numeric, string, and array comparators are now backend-specific classes (PandasNumericComparator, SparkStringComparator, etc.) rather than inline logic. Custom comparators can be injected into any backend compare class.
• validate_tolerance_parameter is now a public API (renamed from _validate_tolerance_parameter).
• Fugue integration removed.
• Python minimum is now 3.12 (CI/tooling); PySpark dependency split by Python version.
• Snowflake: snowflake-snowpark-python minimum bumped to 1.37.

New Features

• Sensitive column masking across all four backends (Pandas, Polars, Spark, Snowflake) — columns can be hidden or hashed without modifying the original DataFrames.
• Custom comparators — pass a comparator instance per column to any backend for fully custom comparison logic.
• cols_with_mismatches method — programmatic access to the set of mismatching column names.
• cache_intermediates option for SparkSQLCompare — controls intermediate DataFrame caching to tune Spark job performance.
• Pandas 3 support.
• Per-column tolerance as a dict in addition to a global float.

Fixes

• SparkSQLCompare: forbid case-sensitive join columns.
• Snowflake: max_diff now correctly defaults to 0 when None.
• Join columns None handling fixed.

Other

• Jinja2 template-based report rendering.
• Copyright year updated to 2026.
• CLAUDE.md added.
• Dependency ranges updated across all backends.