Gavin King

What I'm saying is that: - we should produce `smallfloat` when the requested precision is `InformixDialect.getFloatPrecision()`, and - `float` when it is `InformixDialect.getDoublePrecision()`. Otherwise we will be unnecessarily creating `float`...

Actually perhaps you were merely suggesting to just change that one line of the fix, and leave everything else the same. If that's the case, I agree. It sounded like...

> We could do the same simplification for e.g. PostgreSQL since the parametrizations of float are essentially just aliases as well according to [its documentation](https://www.postgresql.org/docs/current/datatype-numeric.html#:~:text=PostgreSQL%20also%20supports%20the,mean%20double%20precision.). Oh god, please no. I...

Thanks, @VladoKuruc!

You might also want to backport https://github.com/hibernate/hibernate-orm/pull/8420

Hi @mkouba > The built-in implementation is bound to the Servlet API. That might be true, but I'm not sure why it matters? > In theory, you could implement your...

> a user may start multiple conversations concurrently, making things slightly more complicated. @cescoffier so the wrinkle here is: can the "user" start multiple conversations _from within the same request...

> Absolutely. In Quarkus currently we have a major gap in the WebSocket handling which I envision will be a heavily "transport" for such "conversations" And that's even worth solving...

> No matter what was intended I've never seen a custom conversation context impl. Hence it's a theory from my point of view. OK, no problem, now you're (potentially) seeing...

> in reactive messaging - https://github.com/quarkusio/quarkus/issues/21367 here for long time it was not well-defined what `@RequestScope` means See, I mean, I would have said (and I'm pretty sure I _have_...