Github user sihuazhou commented on the issue:
https://github.com/apache/flink/pull/5934
Hi @StefanRRichter , I think I feel lost again... when I writing the
comments for the serializer about why we don't duplicate it, I found a loophole
there. In theory, even tough the serializer is just written, but if the
serializer is stateful (e.g. it maintain a `variable` inside, and do the
serializing works according to it and also may update the `variable` according
to the serialized result) then it might also lead to concurrency problem if we
don't duplicate it. But in practice, I can't find any use cases that need a
stateful serializer as far as I can see... So, now I feel lost in duplicating
it to be on the safer side in theory or just add some comments without
duplicate it to avoid the costs of the duplicating. What do you think?
---
