"Matt Clark" <matt@ymogen.net> writes:
> consider:
> begin;
> update a set col1 = 'p' where id = '1';
> update a set col2 = 'q' where id = '1';
> commit;
> versus:
> update a set col1 = 'p', col2 = 'q' where id = '1';
> Does the first case generate any more dead tuples that will need
> vacuuming than the second case,
Yes. Takes twice as long, too.
regards, tom lane
