Hi Fred, Thanks for the updates!
Questions: K11. Can we reconsider the introduction of two new exception subclasses? Perhaps I don’t understand the benefit? Technically both the key and the value could have deserialization errors, right? K12. Is there a benefit to exposing both the ByteBuffer and byte[]? K13. Isn’t it possible for the key() and/or value() calls to throw NPE? (Or maybe I don’t understand why we need the old constructor :( Thanks, Kirk > On Apr 23, 2024, at 12:45 AM, Frédérik Rouleau > <froul...@confluent.io.INVALID> wrote: > > Hi Andrew, > > A1. I will change the order of arguments to match. > A2 and A3, Yes the KIP is not updated yet as I do not have a wiki account. > So I must rely on some help to do those changes, which add some delay. I > will try to find someone available ASAP. > A4. I had the same thought. Using keyBuffer and valueBuffer for the > constructor seems fine for me. If no one disagrees, I will do that change. > > Thanks, > Fred