Stephen Frost <[EMAIL PROTECTED]> writes:
> Of course, the last time this went around the argument was that we
> shouldn't add alot of extra code until we actually needed to, while at
> the same time we shouldn't use up the few remaining bits we have. The
> fact that this makes for an impossible situation seems to have been
> lost.
No, it hasn't been forgotten at all. Whenever we have to cross that
bridge, we'll do so. The questions being asked here are about whether
an adequate case has been made for adding *user-visible* complexity,
not about nitty little details of internal representation.
There are also some compatibility concerns involved. If we add
grantable privileges for TRUNCATE and/or DDL operations, then GRANT ALL
ON TABLE suddenly conveys a whole lot more privilege than it did before.
This could lead to unpleasant surprises in security-sensitive
operations. One could also put forward the argument that it's a direct
violation of the SQL spec, which after all does specify exactly what
privileges ALL is supposed to grant.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
