Thanks, Kevin.

Well, yes — your point could make sense to me. But never forget Engineering is all about cost-benefit and trade-offs — so as developers/architects we need to make the best decision based on the situation, project, team, and many other factors. But, that’s right — we need to register all of our dependency objects regardless of what they are (command handlers or anything else).

The concern that it could go countless, from my point of view, doesn’t have anything to do with the technique proposed in this article. Rather, that could be more of an architectural concern (code-wise at least). I mean, segregating the domains of your application (Bounded Context) and offering them via different solutions (Visual Studio I mean) could control the growth of registering dependencies (in addition to other advantages) to some extent. Although, taking Microservices kind of approach could bring its own drawbacks too, but again, it’s like a trade-offs game that as software engineers we never get rid of and would learn it better by reading more good codes, experience, and as such.

Good luck