>> Former - on each row application might update different rows and we want
>> to change only them. So the update statement is re-prepared each time.
>
> Looks like something that could be improved by only re-preparing if the
> statement is different.

Yes, but I view those scrollable updatable statements as a convenience
methods for people that use JTree or similar to perform "online" updates
on an active result set from the fat client. Here the optimization will
not bring a lot, since a) people won't notice it; b) it is unlikely that
human will have the same update pattern on many records.

Those, who develop code for the server-side, do not use those methods in
most cases anyway. They either use JPA or other O/R mappers or "talk"
JDBC direct. JPA and O/R mappers do not use these methods either.

So the only target audience are those developers that "talk" JDBC
directly and are lazy to prepare a separate PreparedStatement.

So, now the question - is it worth adding one more A.I.-feature to the
driver and maintain it for that small audience?

Roman