# Changelog for LPPaver

## [v0.0.5.0](https://github.com/rasheedja/LPPaver/compare/v0.0.4.0...v0.0.5.0)
- Use a new heuristic for the best-first search algorithm
  - Centre of averages of the ranges of each term in the conjunction is the heuristic
- Allow larger cutoffs in tests
  - Required for new heuristic
- Remove an unneeded variable when linearising functions

## [v0.0.4.0](https://github.com/rasheedja/LPPaver/compare/v0.0.3.1...v0.0.4.0)

- Rename LPPaver.Decide modules to LPPaver.Algorithm
  - LPPaver.Decide.Algorithm is now LPPaver.Algorithm.DNF
  - Other LPPaver.Decide.Name modules are now LPPaver.Algorithm.Name
- Polish documentation
- Add useful types for our decision algorithms
- Update decision algorithms to use these types, making them easier to read and maintain as well as returning pavings for each algorithm's result
- Add option -o to executable to write pavings to a JSON file.

## [v0.0.3.1](https://github.com/rasheedja/LPPaver/compare/v0.0.3.1...v0.0.3.1)

- Write initial README.md and REFERENCE.md
- Remove haddock documentation using GitHub pages
  - Issues with dependencies so using hackage instead

## [v0.0.3.0](https://github.com/rasheedja/LPPaver/compare/v0.0.3.0...v0.0.2.0)

- Deploy haddock documentation using GitHub pages
- Clean up and document new modules
- Split up single Type.hs module into smaller and better encapsulated modules
- Add constraintRightSide to LPPaver.Constraint.Util
- Document LPPaver.Constraint.Type and LPPaver.Constraint.Util

## [v0.0.2.0](https://github.com/rasheedja/LPPaver/compare/v0.0.2.0...v0.0.1)

- Replace minView with maxView in 'model search' mode
  - Higher ranges are more likely to produce models
- Add test suite
  - Uses PropaFP and 'Place' examples
- Update package dependencies

## [v0.0.1](https://github.com/rasheedja/LPPaver/tree/v0.0.1)

- Initial release