Clojurians Log v2

Is there any guidance out there on organizing larger lacinia projects, particularly in terms of splitting up schema and resolver files? Do people prefer to keep a single large schema file and/or single large resolvers file? And if not, along what lines do they split things up?

At Walmart, we were federated, so we made use of a single SDL file and our architecture involved very few resolvers, primarily just the root operations. Certainly, I vastly prefered the inject-resolvers approach to the attach-resolvers approach (the latter dating back to the non-public Lacinia prototype, before I even joined Walmart).

Interesting, I was wondering about switching to an SDL file from EDN!