I made a fuzz-test thingy

Branch: https://github.com/lydell/prettier/tree/esfuzz

Using: https://github.com/estools/esfuzz/

Command: npm run fuzz.

It generates a random program and then runs it through prettier two times. No errors are supposed to be thrown, and the two outputs from prettier are supposed to be equal.

It prints the following:

<random JS>
─────────────────────────────────────
<prettier JS or parse errors or diff>
─────────────────────────────────────
<random prettier options>
─────────────────────────────────────
<message>

You can easily work with the results if you find something interesting:

• The random JS is written to fuzz/random.js
• The first prettier run-through is written to fuzz/prettier1.js
• The second prettier run-through is written to fuzz/prettier2.js

Example:

Feel free to close this if it is of no use 😃 I’ll play around with it and see if I find anything interesting.

[vjeux]

Awesome! Thanks for all your work on it 😃

Let’s close this issue. Feel free to open up more bugs you find with it!

[vjeux]

This is awesome! Can’t wait to see what you find.

If you look at run_spec.js we already have the infra to make sure that the ast before and after pretty print is the same