Robert,
* Robert Haas (robertmhaas@gmail.com) wrote:
> On Mon, Apr 6, 2009 at 7:30 AM, Stephen Frost <sfrost@snowman.net> wrote:
> > I do such diffs pretty often, but I don't think I've *ever* done it on
> > catalog tables.. Perhaps it'll come up in the future, but I doubt it.
>
> Well the point is when you dump a user table, it will dump this
> setting along with it, same as it does now for statistics_target. So
> if you diff the DDL you might see differences in rounding. If you
> only diff the data, it won't matter unless, as you say, you're dumping
> pg_attribute itself.
The rounding when you dump it out is going to be consistant though, is
it not? I mean, you might get a difference between what you try to set
it to and the result that you get from pg_dump, but if you compare one
pg_dump to another done later there shouldn't be any change, right? If
there is an architecture difference then I could maybe see it, but I
thought float-handling was well-defined on systems we run on.
Thanks,
Stephen