Andrew Dunstan <andrew(at)dunslane(dot)net> writes:
> On 2024-09-12 Th 5:33 PM, Tom Lane wrote:
>> I'm inclined to propose rejecting FK constraints if the comparison
>> operator is not immutable.
> Isn't there an upgrade hazard here? People won't thank us if they can't
> now upgrade their clusters. If we can get around that then +1.
Yeah, they would have to fix the bad DDL before upgrading. It'd
be polite of us to add a pg_upgrade precheck for such cases,
perhaps.
regards, tom lane