Tom Lane wrote:
> Of course; I suppose this bug goes back to Berkeley days. We just
> hadn't recognized it before (or at least I hadn't).
>
Neither had I. But the changes in 7.4 probably make it more likely
people will bump into this as a problem.
Without looking to confirm, I believe SQL99 defines an array as always
having a lower bound of 1, making our behavior an extension to the
standard. We may need another extension to the array literal syntax in
order to deal with this. I'll report back after I've had some time to
study it.
Joe