Bah si les fameuses conditions c'est juste que Macron le traite pas comme une marionnette, y a pas de garantie ça peut toujours changer donc bon...

Non mais sérieux y a aucune source c'est du troll pur à ce niveau là…

Non c'est juste une traduction alternative de "dark energy"

Non c'est le début d'un cri révolutionnaire

The second one

However, if I try to rephrase it: does it mean that the "best" way to identify the hex boundary is that things that are regression tested are inside, and the rest is outside?

I'm sorry but I'm trying to apply the concept to a common (for me) professional environment: a Spring Boot app, so mocks are the way to go. Moreover, I find it easier to understand abstract concepts by starting from a practical example and then generalizing it.

When you say he put the tests "outside" the ACL, you meant the "boundary" of the tests, so the mocks? Does it mean that the inside of the hex is what is tested, and the outside is what is only mocked?

(I mean on the long run, because one strength of this pattern is that temporarily, when switching techs, you can just replace the driven adapters without touching the core and their tests; but after this step, I wonder what should be permanently moved to the core with later refactoring)

Mmmh so does it mean that the driven adapters should stay humble so that you don't have to test them, or that if you "feel" that you have to test some "low level" parts of the "ACL" separately from the "core" then it should be in the driven adapters? (Thanks for the answer and the sources!)

Yeah it makes sense. By the way, it's a little off-topic but do you have any opinion about where AntiCorruption Layers should be? Should the outside of the "core" remain as "humble" as possible, or should the "core" remain as pure as possible, without knowing about legacy systems?

Do you agree that the driving port interface is not really useful in general (except for symmetry and maybe the Interface Segregation Principle) compared to the driven one?