Chris Angelico <rosuav@gmail.com> writes:
> To be quite honest, I would simply read the table directly and then do
> the processing in an application language :) But two window functions
> should do the trick. Whether or not it's actually more efficient that
> way is another question.
FWIW, Postgres is reasonably smart about the case of multiple window
functions with identical window definitions --- once you've got one
lag() in the query, adding more isn't going to cost much.
Having said that, they are pretty expensive. I tend to agree that doing
the processing on the application side might be faster --- but only if
you've got a place to put such code there. If you've just got generic
query-result display code, there may not be any convenient way to do it.
regards, tom lane
