Gavin King

But the correct (portable) behavior is now defined by JPA, and is not something that is supposed to vary between databases. Again, I don't think we need to be messing...

Again: if you're talking about Hibernate 5 here, then yes, correct, there are very many things that are inconsistent between dialects on H5. These have been fixed in Hibernate 6,...

I guess this impacts Hibernate Reactive.

> > I guess this impacts Hibernate Reactive. > > > > I looked at the ReactiveActionQueue and I don't see any cycle detection code in the batch sorter loop....

> > I guess this impacts Hibernate Reactive. > > I looked at the ReactiveActionQueue and I don't see any cycle detection code in the batch sorter loop. @ngavars Could...

> As for the `ReactiveActionQueue` - it will not abandon sorting in this case from what I can tell. Hrm. Thanks. This is a little confusing to me, because the...

This looks like it might impact HR, need to look more closely.

Well if you turn on `hibernate.use_nationalized_character_data` Hibernate _uses_ `setNString()` / `getNString()`, so I think that counts as "not supported", right? I mean, I guess we could put a `catch` in...

> I thought that was exactly the use case for `IMPLICIT`, no? Not exactly, no. With `IMPLICIT` it will use `VARCHAR` and `CHAR` for DDL generation, which on SQL Server...

Well I'm not sure that's quite right either: at least it's quite inconsistent. I think what you're reaching for is a new enumeration value `IMPLICIT_FOR_JDBC_EXPLICIT_FOR_SQL`.