The Boundary (https://github.com/sasa1977/boundary) thing is interesting. It seems logical to want to put dividing lines within a monolith like this, but I'm super curious as to what problems they faced that spurred them to bring it in in the first place? Like, were internal APIs changing so fast that it kept breaking dependants? Wouldn't/shouldn't test coverage help with that? Or did it have more to do with tooling/DX? (e.g. speeding up compile times or something)

Post reply on HN