Niclas Hedhman wrote: > Yes, that makes sense, AND allows for codebase modularity as well, > without cyclic dependencies at that level.
Yup, pretty much. > So, what is that Status now? Are we keeping "Layer" terminology and > don't allow network graphs, or are we scrapping Layer and fall into > 'anything goes', which would then raise the question; We have a > container of Modules, where Modules are 'any to any' but the higher > container is directed graph. Should we have one or both of such > containers, and should that be extended indefinitely. > > Personally, I think the "Layered Architecture" is well understood and > easily accepted, and I think I want to keep it until more overwhelming > evidence of a better metaphor emerges. For now, I agree. Since the hexagonal architecture can be still applied with the "layers as directed graph" rule, it's still "ok", I think. The property of layers that they "use stuff that comes in" and "shows stuff that can be used" still applies, which is useful. In SCA terms it would probably just be a "Component", but that feels a bit too generic. /Rickard _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
