Tests needed

[reggi]

This library needs some form of unit testing. There are over 150 individual functions that make up evalmd, a lot of logging and writing to stdout & stderr, flags, exit codes, relying on package.json, deps that need to be installed. There are two ways of testing evalmd.

1. I can collect full module / repos. Assert an exit code and evaluate the markdown. Something like testSuccess('https://github.com/ljharb/object.assign', 'README.md') (probably wise to specify commit hash) where this would git clone the repo and run npm install then evaluate the markdown. Ideally I can supply optional flags as well. This method can use nixt to ensure the bin itself works as expected, and the exit code is correct.
2. I can test each and every function individually which would be great because them I can ensure code coverage. Or because eval-markdown.js returns one function that utilizes every other function I can just test it, however I still run into an issue where I need — not only a single markdown file to test, but a whole environment with package.json and node_modules.
3. A combination of both methods. (probably wise, however a ton of work)

Thoughts?

[reggi]

@ljharb I'd love your opinion on this. What do you think the best way to test this library is?

[ljharb]

You can test the vast majority by just providing a folder full of example markdown files.

You can then make subdirectories with package.json files and a markdown file to test each of the edge cases of requiring the current package in the example.