I'm not being critical, I just think that in practice we would be
better off porting all of the connect-by statements from the start.
Even with recursion in the db, we've had several complaints about slow
queries that were related to the ported connect-by statements. If we
had something like db_dfs, the temptation would be to not port the
connect-by queries until they became a problem.
The plus side to using db_dfs is that we get a common interface for
fixing the connect-by queries. Since it's possible to do dynamic sql
inside of pltcl functions, I wonder if it wouldn't be possible to
combine db_dfs with a pltcl function that did the recursion inside the
db?
