Neil Conway writes:
> (b) Using the above scheme that attnum == attpos initially, there
> won't be any gaps in the sequence of attpos values. That means
> that if, for example, we want to move the column in position 50
> to position 1, we'll need to change the position's of all the
> columns in positions [1..49] (and suffer the resulting MVCC
> bloat in pg_attribute). Changing the column order is hardly a
> performance critical operation, so that might be acceptable.
I don't think you can speak of "bloat" for pg_attribute. But you can
speak of a problem when you want to do the old col = col + 1 in the
presence of a unique index.
> (c) Do I need to consider inheritance?
Inheritance is based on column names, so initially no, but if there is a
command to alter the column order, then it should have an ONLY option.
--
Peter Eisentraut peter_e@gmx.net