Joe Conway <mail@joeconway.com> writes:
> On 05/31/2013 08:46 PM, Robert Haas wrote:
>> Changing SQL syntax in the back-branches isn't normally something
>> we do, but I confess I don't see any real reason not to do it in
>> this case.
> That was part of my hesitation, but I don't see any better way to fix
> existing installations and this is pretty well self-contained. Any
> other opinions out there?
I don't like this approach much.
1. It does nothing to fix the issue in *existing* databases, which
won't have pg_depend entries like this.
2. In general, we have assumed that properties of tables, such as
indexes and constraints, cannot be independent members of extensions.
It's not clear to me why rules should be different.
regards, tom lane