Peter Eisentraut <peter_e@gmx.net> writes:
> Tom Lane wrote:
>> I believe the reason for
>> not changing it was that it seemed too likely to break existing
>> functions, with potentially nasty consequences if they chanced to be
>> security definers.
> Is this actually true or did we just forget it? :-)
I recall thinking about the point. The decision could've been wrong ...
regards, tom lane