Re: logical changeset generation v6.7

On Tue, Dec 3, 2013 at 8:24 AM, Andres Freund <andres(at)2ndquadrant(dot)com> wrote:
> On 2013-11-28 21:15:18 -0500, Robert Haas wrote:
>> OK, I've committed the patch to adjust the definition of
>> IsSystemRelation()/IsSystemClass() and add
>> IsCatalogRelation()/IsCatalogClass().
>
> Thanks for taking care of this!
>
>> I kibitzed your decision about
>> which function to use in a few places - specifically, I made all of
>> the places that cared about allow_system_table_mods uses the IsSystem
>> functions, and all the places that cared about invalidation messages
>> use the IsCatalog functions. I don't think any of these changes are
>> more cosmetic, but I think it may reduce the chance of errors or
>> inconsistencies in the face of future changes.
>
> Agreed.
>
> Do you think we need to do anything about the
> ERROR: cannot remove dependency on schema pg_catalog because it is a system object
> thingy? Imo the current state is much more consistent than the earlier
> one, but that's still a quite surprising leftover...

I don't feel obliged to change it, but I also don't see a reason not
to clean it up.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company