This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Pending Approval

These branches will be created by Renovate only once you click their checkbox below.

• Update dependency ipython to v8.18.1

Detected dependencies

• Check this box to trigger a request for Renovate to run again on this repository

This will allow local tests to be personalized per user as long as their pytest.ini exists (create via task). Repo tests can be for actual repository logic.

• Have a "Solve day one" article front-and-center, with the below troublesome details linked in, but not necessarily front-and-center
• When installing recommended extensions, it is not necessary to install the "other" recommended extensions
• Give a tour of Sourcery refactoring based on their onboarding materials, specifically differentiate between the refactoring features and the AI features, Opt-in is not required, disable Sourcery Code Lens in settings (consider hard-coding this)
• When setting up .gitconfig on Windows 11 by default, it saves as .txt. Need to enable developer options to turn on file extensions. Cover privacy options l
• Magic terminal activation may not happen before tasks.json task execution. Consider user settings.json experimental flags, recommend user pwsh profile modification, and possibly task rewriting
• Manual breakpointing in notebooks is powerful but confusing. Detail just enough about test harness to convey the point
• Detail the usage of magic notebook objects inp and chk, how these are hijacked for tests, and why you can't just Path("<...>.txt").read_text().
• Detail checkpointing and how one user's solution can help another user solve via a focused test in advent23_tests.parametrize_compare.
• Detail the blasphemous mixture of notebooks next to package modules, why this actually makes sense, and how this isn't a "package" in the traditional sense (not meant to be pip installed from PyPI, meant for refactoring and test/logic isolation, etc.)
• Detail the "basic anatomy" of the repo, leaving details out for deeper write-up later
• Detail interactions with Ruff and Pyright, especially the meaning of warnings/highlights and their severity
• Detail the necessary evil of dealing with a flaky Problems pane
• When suppressing Sourcery, make sure you do it locally

• Copy notebooks over at build time, properly .gitignored
• Run in documentation build separately
• Generate toctrees automatically
• Isolate from main documentation

Get attempted checks for a notebook like in Attempt.checks, but also return the nodes of the right-hand-side of assignments. Now get source code snippets given the original source and replace the RHS in code with values from the chk namespace obtained from executing the notebook.

Remove all lines not starting with chk in code cells after the first.

Produce template notebooks in this fashion.

• Move input handling from advent23_tests to advent23 as e.g. get_input (consider name similarities)
• Source advent23_tests.Case.inp and advent23.get_inp from it
• Update get_inp to allow custom user inputs, e.g. yours or other user input

def get_inp(day: int, user: str = ""): ...

The while result != ... will misbehave because $$ turns to $, so a bare substitution will be necessary on the LHS, e.g. while Template(result). substitute() !=, or something similar. May need safe_substitute` or something like that.

Consider that $$ may enter back into the result pattern at any time, therefore this cannot be handled prior to the loop.

An is_valid() check or straight up $$ in ... may also work.

Clean up old files as well, like input/ex_*. Point to new input/examples.toml as the place to put inputs and checkpoints. Show how to update those.

Don't run notebooks to discover checks, instead analyze the code nodes using nbformat and ast for keys of chk. Also, determine whether checks can be skipped if they are hardcoded (e.g. not set to a variable name).