First, an attempt to provide docstrings to the repository was disastrous. The documentation makes very little sense in isolation, and there is no top-level documentation that pulls it all together.
Second, the code is rife with poor programming practices, like single-letter variable names, which only exacerbate the first issue.
The remediation of this issue will likely span multiple changes. It's possible that the best outcome is a refactor towards typing (as opposed to the nested dict/tuple kerfuffle that exists today. The best outcome is not a requirement for closing this issue.
Adapt has two core readability problems:
First, an attempt to provide docstrings to the repository was disastrous. The documentation makes very little sense in isolation, and there is no top-level documentation that pulls it all together.
Second, the code is rife with poor programming practices, like single-letter variable names, which only exacerbate the first issue.
The remediation of this issue will likely span multiple changes. It's possible that the best outcome is a refactor towards typing (as opposed to the nested dict/tuple kerfuffle that exists today. The best outcome is not a requirement for closing this issue.