-
Notifications
You must be signed in to change notification settings - Fork 4
Droop
Jonathan Lundell edited this page May 14, 2020
·
2 revisions
Droop
is a Python-based package for counting STV elections.
The primary motivation for Droop
is to provide a framework for implementing STV counters in a manner that facilitates verification and certification. Rule implementations are self-contained and independent of other rules. Rules can, in general, be expressed in terms of their statutory language. Extensive and extensible unit tests provide additional validation support.
Droop
's features include:
- STV rule implementations are self-contained, for clarity and straightforward validation
-
Droop
's framework provides housekeeping support tailored for STV implementation, keeping the rule implementations uncluttered. - The framework includes interchangeable arithmetic support for integer, decimal fixed-point, guarded decimal fixed-point (quasi-exact) and rational (exact) arithmetic.
-
Droop
includes implementations of several STV rules, including reference rules from the Proportional Representation Foundation and an example implementation of Minneapolis MN's STV rule; more are planned. - Parametric implementations of Meek, Warren and Weighted Inclusive Gregory (WIGM) methods are available, so that ad hoc rules can be easily assembled and compared from the user interface.
- New rule implementations can be "dropped in" without any changes to the framework.
- The framework, rule implementations and user interface are decoupled; not only are new rule implementations easy to add, but so are new user interfaces. The current implementation includes a general-purpose CLI as well as examples of specialized CLI interfaces for specific rules. A web-based UI is planned but not yet implemented.
- The framework and rule implementations are readily embedded as an STV counting engine in third-party software.
- All IO treated as UTF-8.
See DroopArchitecture for an overview of Droop
's structure.
Currently (v0.14), the unit tests run correctly on macOS 10.15.4 (with Python 3.7.3).