"David G. Johnston" <david.g.johnston@gmail.com> writes:
> ... âBut since ISO 8601 doesn't define infinity our implementation of this
> behavior is not externally constrained and probably shouldn't be changed.
But we already did change it, ie, 9.4 is failing to follow the precedent
established by 9.2 and 9.3. And that was not an intentional thing, nor
is it more reasonable than the previous behavior, so I think it's a bug.
regards, tom lane