Robert Haas <robertmhaas@gmail.com> wrote:
> Well, Tom and I are on opposite sides of this, I suppose. I
> prefer ERROR for everything other than the top-level transaction
> commands, and see no benefit from opting for a wishy-washy
> warning.
+1
If the user issued a local command outside of a transaction there
is an extremely high probability that they intended to either set
session state or affect the next transaction. Either way,
modifying the database with the wrong setting could lead to data
corruption -- at least for some of these settings. IMO it would be
foolish to insist on consistency with prior releases rather than on
giving people the best shot at preventing, or at least promptly
identifying the cause of, data corruption.
Obviously, changing the level of these is not material for back-
patching.
--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company