On Thu, 2003-06-05 at 01:53, Tom Lane wrote:
> Rod Taylor <rbt@rbt.ca> writes:
> > [above]
>
> Er, weren't you just arguing for the opposite?
Yes, I was. Until I realized that some things in pg_dump depend on the
specific name of the constraint in cases where they are auto-generated.
To fix the problem, they need to be common.
Someday I'll probably propose that cluster should work on a list of
columns, and build the index by itself if one is not already available.
Once thats done, we can probably revisit pg_dump and make these items
unnamed in the dump file.
--
Rod Taylor <rbt@rbt.ca>
PGP Key: http://www.rbt.ca/rbtpub.asc