Andrew Dunstan <andrew(at)dunslane(dot)net> writes:
> Tom Lane wrote:
>> That would work. You'd need two state flags instead of just one, but
>> that doesn't seem bad.
> 'splain please :-)
Maybe you weren't thinking of the same thing, but what I was imagining
was one state flag to remember that you'd created the interpreter (and
loaded the unsafe-func support into it), then a second one to remember
whether you've loaded the safe-func support. There are various ways to
represent this of course, but the point is there need to be three
persistent states.
regards, tom lane