Alvaro Herrera <alvherre@dcc.uchile.cl> writes:
> On Mon, Jan 31, 2005 at 12:29:12PM -0500, Tom Lane wrote:
>> Of course, sooner or later we should fix the underlying locking
>> mechanism so we don't have to have this constraint, but I'm not sure
>> when that will happen.
> What needs to be locked?
Somehow we'd need to prevent any changes to the template database while
it's being copied. AFAICS this would require something on the order of
taking out a lock at the database level for every query, which sounds
like a mighty expensive solution compared to the seriousness of the
problem.
regards, tom lane