Skip to content

CHANGELOG.md

Latest commit

 

History

History
199 lines (154 loc) · 9 KB

CHANGELOG.md

File metadata and controls

199 lines (154 loc) · 9 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

[0.3.0] - 2025-12-24

Changed - BREAKING CHANGES

Dependency Management Overhaul: Consolidated 11 dependency files into single-source-of-truth system using pyproject.toml with pip-tools lockfiles.

  • REMOVED: requirements-dev.txt (replaced by requirements-dev.lock)
  • REMOVED: scripts/freeze_requirements.py (replaced by pip-compile)
  • REMOVED: scripts/generate_docs_requirements.py (replaced by pip-compile --extra=docs)
  • Migration required: See docs/MIGRATION-DEPENDENCY-OVERHAUL.md

Developer Workflow Changes:

# OLD (v0.2.x):
pip install -r requirements-dev.txt

# NEW (v0.3.0+):
pip install -r requirements-dev.lock

Dependency Updates: Minimum versions updated for NumPy 2.0 ecosystem compatibility

  • numpy: >=1.22,<2.0 → >=1.26,<3.0 (adds NumPy 2.0 support)
  • scipy: >=1.10 → >=1.13
  • pandas: >=1.5 → >=2.0
  • numba: >=0.57 → >=0.59
  • docstring-inheritance: >=2.0 → >=2.2.0,<3.0 (MRO fix, exclude breaking v3.0)
  • pytest: >=7.4.4 → >=8.0
  • pytest-cov: >=4.1.0 → >=6.0

Added

  • Lockfiles for reproducible builds:

    • requirements.txt - Production dependencies (from [project.dependencies])
    • requirements-dev.lock - Development dependencies (from [project.optional-dependencies.dev])
    • docs/requirements.txt - Documentation dependencies (from [project.optional-dependencies.docs])

  • Tests: tests/fitfunctions/test_metaclass_compatibility.py

    • Validates FitFunctionMeta MRO compatibility with NumpyDocstringInheritanceMeta and ABCMeta
    • Prevents metaclass regression bugs
    • Tests abstract method enforcement, docstring inheritance, all fitfunction instantiation
    • Includes version constraint validation (docstring-inheritance >=2.2.0,<3.0)

  • Documentation: Comprehensive migration guide at docs/MIGRATION-DEPENDENCY-OVERHAUL.md

    • Breaking changes overview
    • Old vs new developer workflows
    • NumPy 2.0 compatibility matrix
    • CI/CD changes
    • Rollback procedures
    • FAQ with common migration questions

  • Dependency Groups in pyproject.toml:

    • [project.optional-dependencies.test] - Testing tools only
    • [project.optional-dependencies.docs] - Documentation tools only
    • [project.optional-dependencies.dev] - All development tools (test + docs + dev)

Fixed

  • Critical: numpy==2.2.6 in requirements.txt violated pyproject.toml constraint <2.0
    • Root cause: freeze_requirements.py used pip freeze without validating pyproject.toml
    • Fix: Replaced custom scripts with pip-compile which enforces constraints
  • Dependency fragmentation: Eliminated sync issues between 11 dependency files
  • Version drift: Lockfiles prevent undocumented version changes

Infrastructure

GitHub Actions: All CI/CD workflows updated for lockfile-based dependency management

  • .github/workflows/sync-requirements.yml:

    • Triggers on pyproject.toml changes (single source of truth)
    • Uses pip-compile to generate lockfiles instead of Python scripts
    • Validates lockfiles with pip install --dry-run

  • .github/workflows/continuous-integration.yml:

    • Uses requirements-dev.lock instead of requirements-dev.txt
    • Faster caching via lockfile hash
    • Cross-platform testing: ubuntu/macos × Python 3.11/3.13

  • .github/workflows/ci-master.yml:

    • Updated to use requirements-dev.lock
    • Consistent with other workflows

  • .github/workflows/security.yml:

    • Audits requirements-dev.lock with safety and pip-audit
    • Security scans on frozen versions instead of loose constraints

  • .github/workflows/publish.yml:

    • Pre-release validation: Blocks PyPI deployment if lockfiles are out of sync with pyproject.toml
    • Prevents releasing with inconsistent dependencies

Scripts: Updated scripts/requirements_to_conda_env.py

  • Now reads lockfiles (default: requirements.txt) instead of requirements-dev.txt
  • Documentation clarifies pip-compile is a prerequisite
  • Supports generating conda environments from any lockfile

Testing

  • NumPy Compatibility: Validated with NumPy 1.26.4 and 2.2.6 (247 tests passed each)
  • Coverage: Maintained 78% (improved from 77.86% baseline)
  • Test Suite: 1576 tests passed, 19 skipped
  • Metaclass Tests: 9 new regression tests for FitFunctionMeta MRO compatibility

Migration

For Developers:

  1. Update checkout: git pull
  2. Install from lockfile: pip install -r requirements-dev.lock
  3. Verify: pytest -q

For CI/CD Pipelines:

  • Replace pip install -r requirements-dev.txt with pip install -r requirements-dev.lock

Rollback: Use pip install solarwindpy==0.2.0 if issues arise

See docs/MIGRATION-DEPENDENCY-OVERHAUL.md for complete migration instructions

[0.2.0] - 2025-11-12

Changed

  • BREAKING: Minimum Python version raised from 3.10 to 3.11
    • Aligns with scientific Python ecosystem (NumPy 2.x, Astropy 7.x require Python 3.11+)
    • Python 3.10 reaches end-of-life in October 2026
    • Enables Python 3.11+ performance improvements (10-60% faster in many workloads)
    • Added Python 3.13 to CI testing matrix for forward compatibility

Fixed

  • Resolved conda-forge feedstock Issue #8 (Python version compatibility)
  • Removed all Python 3.10 references from CI and packaging configuration
  • Updated ReadTheDocs configuration to use Python 3.11

Added

  • Python 3.13 CI testing for forward compatibility validation
  • Runnable Quick Start example in README with realistic solar wind data
    • Demonstrates complete Plasma object creation workflow
    • Includes physically accurate parameter values
    • Users can copy-paste and execute immediately

Documentation

  • Updated installation requirements in README.rst and docs/source/installation.rst
  • Fixed LICENSE file detection for GitHub (converted from .rst to plain text)
  • Archived completed documentation to reduce AI context overhead

[0.1.5] - 2025-11-10

Fixed

  • Documentation validation - Resolved doctest failures for JOSS submission
    • Added continuation markers (...) to multi-line doctest examples
    • Completed Ion class example with all required columns (v.x, v.y, v.z, w.par, w.per)
    • Added # doctest: +SKIP directives to non-deterministic fitfunction examples
    • Added # doctest: +NORMALIZE_WHITESPACE for pandas DataFrame output
    • All 33 doctests now passing (11 executed, 22 appropriately skipped)
    • Aligns with paper statement: "fitfunctions tests remain in active development"
    • Unit tests (1,557 test cases) provide comprehensive functionality validation

Changed

  • Documentation examples - Maintain instructional value while ensuring reliable validation
  • JOSS paper - Updated acknowledgements to reflect AI-assisted development workflow
  • Conda channels - Switched to conda-forge only (removed Anaconda defaults channel)
    • Eliminates commercial channel licensing warnings in CI
    • All dependencies available on open-source conda-forge channel
    • Users with existing environments should recreate: conda env remove -n solarwindpy && conda env create -f solarwindpy.yml
    • Aligns with JOSS open-source infrastructure requirements

[0.1.0] - 2025-08-23

Added

  • Initial stable release on PyPI - First public release of SolarWindPy
  • Semantic versioning with setuptools_scm - Automatic version detection from git tags
  • Automated deployment pipeline via GitHub Actions - Complete CI/CD for PyPI publishing
  • Core plasma physics calculations and data structures - Multi-species plasma analysis
  • Plotting and visualization capabilities - Publication-quality scientific plots
  • Instability analysis tools - Solar wind plasma instability calculations
  • Comprehensive test coverage - ≥95% code coverage with scientific validation
  • Release automation scripts - check_release_ready.py and bump_version.py tools
  • Comprehensive documentation - Release process and deployment guides

Changed

  • Migrated from development to stable release - Production-ready package
  • Enhanced package metadata for PyPI distribution - Complete project configuration
  • Improved version detection with setuptools_scm - Tag-based versioning
  • Enhanced GitHub Actions workflows - Production deployment automation

Fixed

  • Graceful handling of missing PyPI tokens - Continues deployment without tokens
  • Code formatting standardization - Black formatting across entire codebase
  • Test fixture scope issues - Module-level fixtures for cross-class access
  • Documentation validation - Comprehensive doctest and example validation

Security

  • Added validation gates for release deployment - Multi-stage verification process
  • PyPI token security - Secure repository secrets management