Ryan Kelly <rpkelly22@gmail.com> writes:
> I'm having trouble understanding why it is necessary to generate a new
> tuple even when nothing has changed. It seems that the OP understands
> that MVCC is at work, but is questioning why this exact behavior occurs.
> I too have the same question.
It's not *necessary* to do so. However, avoiding it would require
sitting there and comparing the old and new tuples, which would be a
waste of cycles for most applications, which don't do useless updates.
We'd be trading off a small loss on every update for a moderate gain
on a few updates, with all of the benefit going to poorly-coded
applications. This has come up before and we've always judged that
it would be a net loss not gain of performance to check for useless
updates.
regards, tom lane