On Fri, 2007-11-09 at 04:05 -0500, Tom Lane wrote: > Simon Riggs <[EMAIL PROTECTED]> writes: > > I had understood the discussion to conclude that indexes that do not > > depend on any column of the table to not be allowed at all. > > That was my first reaction too, but the point about unique-index behavior > refutes it. Constraining a table to have at most one row is useful.
Sure is, and I've done it just a few days ago. This SQL does it using standard syntax: create table foo (handle integer primary key check (handle = 1)); It's also a lot more obvious than creating an index on a constant, which seems like a wrinkle that we should disallow. It just sounds to me like something that will break again in the future, so I'd rather disallow it now rather than here "but we need to support zero column indexes too". -- Simon Riggs 2ndQuadrant http://www.2ndQuadrant.com ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match
