On 5/25/17 09:55, Robert Haas wrote:
> On Thu, May 25, 2017 at 8:32 AM, Peter Eisentraut
> <peter.eisentraut@2ndquadrant.com> wrote:
>>> Well, I think if it's not going to work, it should be prohibited,
>>> rather than seeming to work but then not actually working.
>>
>> Here is a similar case that pg_dump fails on:
>>
>> create table information_schema.test1 (a int);
>> create view public.test2 as select * from information_schema.test1;
>>
>> It's not clear how to address that, or whether it's worth it.
>
> Sure, that case is hard to address. But why is *this* case hard to address?
They are the same cases.
a) Create object in information_schema.
b) Create another object elsewhere that depends on it.
c) pg_dump will dump (b) but not (a).
So the fix, if any, would be to prevent (a), or prevent (b), or fix (c).
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
