Re: dealing with extension dependencies that aren't quite 'e'

At 2016-04-04 18:55:03 -0300, alvherre@2ndquadrant.com wrote:
>
> At this point I think we're missing user-level docs for the additional
> clause in each ALTER command.

Thanks for having a look. Now that you're happy with the grammar, I'll
write the remaining docs and resubmit the patch later today.

> Do you have any ideas on how this would appear in regression tests?

No specific ideas.

At a high level, I think we should install an empty extension, create
one of each kind of object, alter them to depend on the extension, check
that pg_depend has the right 'x' rows, then drop the extension and make
sure the objects have gone away.

Does that sound reasonable? Suggestions welcome.

-- Abhijit