On sön, 2011-07-03 at 17:02 -0400, Tom Lane wrote:
> Yeah. If there were One True Way to create a UUID, I would probably
> agree that we should push that functionality into core. But there are
> a lot of ways (and the reason for that is that they all suck in one
> fashion or another :-(). Between that and the lack of portability of
> many of the better ways, this is something I'm happy to keep at arm's
> length.
There is a standard, and it defines four ways to create UUIDs. AFAIR,
no one has ever claimed that there is another recognized way that we
don't support. So I don't think that's really the argument. The
argument was that we don't want to bother.