Re: Change pfree to accept NULL argument

On Wed, 24 Aug 2022 at 23:07, David Rowley <dgrowleyml(at)gmail(dot)com> wrote:
> One counter argument to that is for cases like list_free_deep().
> Right now if I'm not mistaken there's a bug (which I just noticed) in
> list_free_private() that would trigger if you have a List of Lists and
> one of the inner Lists is NIL. The code in list_free_private() just
> seems to go off and pfree() whatever is stored in the element, which I
> think would crash if it found a NIL List. If pfree() was to handle
> NULLs at least that wouldn't have been a crash, but in reality, we
> should probably fix that with recursion if we detect the element IsA
> List type. If we don't use recursion, then the "free" does not seem
> very "deep". (Or maybe it's too late to make it go deeper as it might
> break existing code.)

Hmm, that was a false alarm. It seems list_free_deep() can't really
handle freeing sublists as the list elements might be non-Node types,
which of course have no node tag, so we can't check for sub-Lists.

David