Shouldn't side-effect-cache set to false by default in CI environments?

[ayZagen]

I spent hours to find why bcrypt fails in container built from CI but not from my local machine. TIL (correct me if i am wrong pls.), pnpm doesn't execute postinstall scripts for cached dependencies. Even pnpm rebuild bcrypt was using cached ones.

Setting side-effect-cache=false in .npmrc resolved the issue. But i think pnpm should behave like this by default in CI environments.

There could be several things would result in a faulty containers, such as using distributed caches with a different arch/os, multi-layered builds etc. In this situation the error was discovered in runtime, meaning build stages passed successfully despite of it was faulty.

What do you think?

@zkochan I didn't open an issue for this but I really would like to hear your thoughts about it.