On Wed, 21 Jul 2004, Markus Bertheau wrote:
> shouldn't it be illegal for an immutable function to call a stable one?
It's expected that the function's creator is responsible for properly
marking its stability. This allows some flexibility when you know more
than the system does (for example, a stable function that isn't immutable
because of its use of a GUC variable being called from a function that
sets said variable).