You've got an npm package that exposes a CLI, we'll call it sndwch and accepts some args:
sndwch --topping ham --topping cheese
But then someone decides they want to run that with node's --inspect-brk flag (or maybe --require flag, or any other node-based flag). So they have to do this:
node --inspect-brk ./node_modules/.bin/sndwch --topping ham --topping cheese
The problem is that doesn't work on windows, so now you have to do this:
node --inspect-brk ./node_modules/sndwch/bin.js --topping ham --topping cheese
But now the location of your binary file in your package is part of the API and if you ever decide to change how that works it's technically a breaking change.
Solution:
People can install this package and then do this instead:
binode --inspect-brk -- sndwch --topping ham --topping cheese
Sometimes a package has a different binary name from the package name (like which calls it's binary node-which). There's no way to do a reverse-lookup of a binary to find the package for it, so you have to supply that by providing the package name along with the executable you want. Something like this:
binode --require ./mocks -- @remix-run/dev:remix dev
That's it..
Here's everything you need to know:
Everything on the left side of the -- is flags for node
The first thing after the -- is the binary you want to use
Everything after the binary you want to use is flags for that binary
If the package name and binary name are different, you must supply both after the -- by separating them by a : (for example: pkgName:binName).