Gavin King

> It makes sense to mark `Nullable` and `Nonnull` annotations as deprecated, in favor of JSpecify annotations, as they are only to be used in IDEs and build tools, other...

> currently these annotations in Jakarta Common Annotations lack tools support(IDE, and build tools) like JSpecify. That's not really true; they work perfectly well in IntelliJ.

@yrodiere Why are you writing: ```sql SELECT document.id, document.title, document?.documentType?.name, document?.author?.name, document?.author?.manager?.name, document?.folder?.name FROM Document document ``` I would have expected this to be: ```sql SELECT document.id, document.title, document.documentType?.name, document.author?.name,...

Another possibility, which has the (arguable) advantage of (arguably) fitting in more naturally with the aesthetics of SQL would be something like this: ```hql SELECT document.id, document.title, OUTER document.documentType.name, OUTER...

The issue here is that an entity in the **persistent** state _must_ have a well-defined persistent identity. Yes, right from the very moment that `persist()` is called. But in these...

The name is just a suggestion, but I chose it based on the following reasoning: 1. `StatelessEntityManager` is (a) verbose, (b) ambiguous (is it a manager for stateless entities?), and...

> Should ` T getReference(Class, Object id)` be here? I was about to answer "no", but then I realized it does make sense in combination with `fetch()` and for setting...

@quaff I don't care much. I agree that returning the id is not necessary, but sometimes it might be convenient.

I've now made a good start on the spec side of this. I think I've got the impact on chapter 3 pretty much nailed. This change will also have an...

I guess we need analogs of `runInTransaction()` and `callInTransaction()`. I'm deeply offended by the naming asymmetry of `runWithAgent()` and `callWithAgent()`, but it's the best I can think of right now...