In short: domain roles belong where the domain lives. Putting them on the auth server couples it to someone else’s business model, and baking them into the JWT freezes them for the lifetime of the ...
In short: input validation is a filter, not a guarantee. It rejects malformed requests before they cost you anything, but it can’t see what’s already in the table. Uniqueness and integrity are fact...
In short: idempotency isn’t about rejecting duplicate requests — it’s about making sure the same intent, delivered more than once, never triggers the side effect a second time. A retry after a drop...
In short: a cache can decide the outcome of an operation before the database is ever asked. Checking a cache before a write looks like a pure performance trick — skip the redundant query, save a ro...
In short: a transaction boundary isn’t determined by which SQL queries physically execute together, but by which component knows about the business invariant that ties multiple repository calls tog...
In short: an aggregate is worth loading only when its state is required to decide whether the new state is valid. If an operation is fully determined by its input and doesn’t check any domain invar...
The relational model stores the FK on the child table side. In Java we have two ways to reflect this relationship: a collection in the parent entity (@OneToMany / List<Child>) or a reference ...
1. Context and Scope The question “where should validation live” is usually reduced to a dispute between Fat Database, Transaction Script, and Rich Domain Model. In practice this is the wrong fram...
The use-case contract should describe what the application needs — not the API of the persistence framework it happens to use. Spring Data pagination works well in CRUD applications. For many proj...
JPA does not provide a first-class model for partial nested graphs as a concept. For that you need JDBC (manual assembly), jOOQ (MULTISET), or Blaze Persistence (Entity Views). Most discussions ar...