todos.org

Graphsat futuristic improvements

• Read: 12 steps to scientificpython : http://www.davidketcheson.info/2015/05/10/rock_solid_code.html
• Try out concurrency for apply_patterns.py as well as pattern_recognition.py helpful link
• Make it a Sage package.
• Python Package Structure: Dead Simple Python: Project Structure and Imports - DEV https://dev.to/codemouse92/dead-simple-python-project-structure-and-imports-38c6

Implement rule rewriting using PLY (Python Lex-Yacc)

docs

http://dabeaz.com/ply/

github

https://github.com/dabeaz/ply

Construct the following graph-satchecker

Take a graph. Compute all Cnfs on it. Translate each CNF by the correct amount so that each CNF has a disjoint variable set. Then send this big-bloated CNF to a cnf-satchecker. Question: Is this strategy faster than other graph-satcheckers?

Use hypothesis for property-based testing of graphsat functions

Make UML diagrams for all of graphsat using PlantML

The multi dispatch module will make graphsat better

https://pypi.org/project/multipledispatch/#description

Post it on PyPI

Add an inventory of each file, it’s purpose, it’s format.

This should include the data files and config files.

Add instructions on how to carry out all the calculations from the paper/thesis

Add a clear pointer to all the project dependencies.

How to keep these dependencies up to date? Mamba perhaps? If we decided to use a pyproject.toml then this PEP is a guide to writing that file: https://www.python.org/dev/peps/pep-0518/

Add installation instructions for both conda and pip.

Have raw → intermediate → final data available for download

Add contact info: in case user has questions/comments

But for issues they should open a new issue on github.

Add instructions on how to cite the software

Create a docker container image

Perhaps a VirtualBox image for getting the running environment quickly

Public outreach: add visuals and examples and other popular things for layperson