ROADMAP.md

BEAST Road Map

Stage 1: Bytecode Interpreter (Virtual Machine)

• Build the general pipeline for defining programs and running them
• Cover the entire code base
• Add more operators to have a solid base of operators available
• Write up documentation for how things work, what the architecture is, etc.
• Set up CI/CD using CircleCI, GitHub CodeQL, AppVeyor
• Set up automatic build of Doxygen and Sphinx documentation using the ReadTheDocs service
• Set up GitHub repository with all contributing guidelines, code of conduct, security guidelines, license, issue templates, etc.
• Clean up all code and documentation, finish documenting all entities
• Add more examples for how to use the virtual machine and implemented operators

Stage 2: Genetic Algorithms and Program Evaluation

• Develop program factories that semi-randomly generate programs
• Develop runtime statistics protocols that record specific execution characteristics of programs
• Develop program performance evaluators, taking into account runtime statistics, but also specific criteria that programs are supposed to satisfy, and scoring them
• Develop a genetic algorithms based approach to recombining programs that perform well based on the implemented criteria

Stage 3: High level Language

• Develop a high level language that allows users to write BEAST programs in a script language
• Consider using Boost Spirit for parsing the high level language

Stage 4: GPU Support

• Add GPU support for executing BEAST programs

Future Stages

• TBD