John Worsley <lx(at)openvein(dot)com> writes:
> Yeah, that's a cleaner solution. I take it anything pstrdup'd by
> PostgreSQL gets freed automatically by the backend?
Pretty much. The only situation where it wouldn't be is if
CurrentMemoryContext is pointing at TopMemoryContext or another
long-lived context --- but we are *very* chary about how much code we
allow to run with such a setting. User-definable functions can safely
assume that palloc'd space will live only long enough for them to return
something to their caller in it.
regards, tom lane