Re: MOVE strangeness

Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > Why does the MOVE -3 return 2?
>
> Because he's successfully backed up over 2 real rows. Had he done FETCH
> -3 in the same situation, he'd have gotten back 2 rows; there is no
> third row it could have returned, so it's hard to argue that the count
> should be anything but 2. (If you think it should be 3, what if I say
> MOVE -10000? Should I get back a count of 10000?)

Sorry, I am not understanding. If he does:

jtv=> begin;
BEGIN
jtv=> declare c cursor for select * from events;
DECLARE CURSOR
jtv=> fetch 3 in c;
year | event
------+------------------
2010 | A Space Oddyssey
2010 | Oddyssey Two
2038 | time_t overflow
(3 rows)

here, isn't he sitting at the start of the fourth row, no? When he does
MOVE -3, doesn't he move from the start of the 4th row to the start of
the 1st row?

jtv=> move -3 in c;
MOVE 2

> > If he has fetched 3, he is at the end of
> > the third row. If he does MOVE -3, hasn't he moved backward three rows
> > to the start of the first row?
>
> There is no "end of a row" as distinct from "start of a row". You can
> be on a row, or before the first row, or after the last row. There is
> no other state besides that.

So, you are saying if he does a FETCH 3, he is still sitting on the 3rd
row, rather than at the gap between the 3rd and 4th rows? If so, the
behavior does make sense.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073