Ongoing project list

[david-yz-liu]

This is a placeholder for a list of ongoing project ideas, to be split into separate issues as people work on them.

• Pylint message improvements. Over the break, I went through all of the existing Pylint checks and made notes about which ones were currently undocumented, and which ones I wanted to improve the error message for. Either or both of you could work on these improvements incrementally (this is nice because the tasks are already divided up cleanly).

• Runtime contract checking. This is a new feature that I'm hoping to deploy in Fall 2020. You can see the prototype under python_ta/contracts; the idea is similar to what you did for the last part of CSC324 A1. The prototype works but isn't very robust, and needs to be extended to support a bunch of different things. Because this is such a new feature, this is also an opportunity for some creative brainstorming/design. I think one or both of you could work on this, but I don't know there's enough work to last the whole semester (which is okay!).

• Investigate incorporating mypy (static type checker). mypy is a static type checker for Python that's actively being developed. I've had students work on a static type checker as well (see python_ta/typecheck) but the work is incomplete. I'm leaning towards switching to mypy, but the downside is I think it's a fairly closed API, so I'm not sure how much control/configuration we'd be able to do. But it also supports plugins, so it could actually be quite configurable.

• Investigating astroid "inference" and constant propagation. This is sort of a hybrid between static and dynamic analysis. Basically, constant propagation is a static analysis that determines values of variables that are just constants, e.g. in

  x = 1
  y = x + 3  # y = 4
  z = x + y  # z = 5
  

  This could be useful as a generalization of our "constant_test_checker", which is a mistake students sometimes might make. Astroid's inference might do some/all of this already, but we could also use control flow graphs I think.

• Data collection. Basically, we can implement a mode for PyTA so that every time it runs check_all, it sends the code being inspected to a server somewhere, along with the configuration used to run pyta. Eventually, instructors would be able to use this data to analyze how students are using PyTA. Obviously, this would incorporate a permission setting on individual installations and a server back-end to receive and store the files.

• Command-line interface. Another more general technical task, but it would be nice to be able to run python_ta on the command-line, like pylint itself. This would involve learning about how to turn Python modules into command-line executables.

• PyCharm plugin integration. It would be nice to have a plugin for PythonTA and PyCharm. This again would require some pretty open-ended investigation of how PyCharm plugins work.