understanding module boundary enforcement

[epikhighs]

Hi, I asked this question on stack overflow as well, but thought I'd try to raise more awareness here too. My motivation behind this discussion is from a position of ignorance and I'd like to correct that ignorance by understanding a bit more than I currently know.

So I'd like to know for NX libs, why module boundaries are enforced through a public API (everything that gets exported from the libraries' index.js). I can understand the general benefits of proper modules boundaries such as:

1. better organization through colocation or similar file location
2. possibly easier for code splitting

But to attain these general benefits, I'm not sure if module boundaries must be enforced through a public API. For larger teams and monorepos that contain multi-platform projects or multi-purposes projects, I can see more strict control might be necessary, but for smaller teams and with more single-purpose projects, I'm not sure if as strict control is needed.

To be clear, the "control" I'm specifically referring to is the eslint rule nx-enforce-module-boundaries and requiring public APIs only be exposed through index.js instead of importing files/methods directly (refer to the code sample in the SO question)