Robert Haas <robertmhaas@gmail.com> writes:
> On Tue, Apr 5, 2011 at 6:03 AM, Shigeru HANADA
> <hanada@metrosystems.co.jp> wrote:
>> * The comment_user_mapping_core.patch includes syntax support, catalog
>> manipulation, pg_dump support, documents and regression tests.
> I don't think it's going to fly to add a function
> pg_usermapping_ownercheck() with a randomly different API than all the
> parallel functions for other object types. There is probably some
> more refactoring that needs to be done here to make this sane, but I'm
> coming around to the view that trying to slip this into 9.1 is not the
> best thing for us to be spending time on, especially considering that
> it doesn't seem to be straightforward to figure out how it should
> actually work. I am inclined to punt this to 9.2.
I agree --- this can clearly contains more worms than we expected.
Supporting user mappings in COMMENT, EXTENSION, etc is not so critical
that we should push a possibly misdesigned notion of ownership into
the system for it. Better to take our time and think about that.
(BTW, it might be useful to reconsider casts while we are thinking about
this. Those don't have a proper notion of ownership either. I'm a bit
inclined to think that we should just bite the bullet and add owner
columns to both these catalogs. But, again, let's not be hasty.)
regards, tom lane