Rob Hasselbaum Sorry to trudge up such an old issue, but I am working on HHH-9808 where read/write fragments seem to have been applied inconsistently for identifiers (pks). Did we have some discussion about this in particular?

E.g. the write fragment is not accounted for at all. Also, in terms of SQL generation the read fragment is applied sometimes, other times not depending on how the SQL fragment is generated.