On tor, 2010-01-28 at 10:34 -0500, Tom Lane wrote:
> Heikki Linnakangas <heikki.linnakangas@enterprisedb.com> writes:
> > ISTM you should explicitly grab a lock on the of-type at some point, to
> > make sure it doesn't get dropped while you're busy creating the table.
> > How do we protect against that for the types used in columns?
>
> We don't. There is no concept of a lock on a type.
>
> For scalar types this is more or less irrelevant anyway, since a scalar
> has no substructure that can be altered in any interesting way. I'm not
> sure how hard we ought to work on making composites behave differently.
> I think it's as likely to cause problems as solve them.
The right thing would probably be SELECT FOR SHARE on the pg_type row,
but I don't see that sort of thing used anywhere else in system catalog
changes.