On Fri, Oct 16, 2009 at 01:28:57PM -0400, Tom Lane wrote:
> David Fetter writes:
> > I think what the OP was expecting was to have the last value be 3
> > instead of changing. This is at least a POLA violation.
>
> [ shrug... ] It's what the spec requires, as far as anybody here
> can tell. As
David Fetter writes:
> I think what the OP was expecting was to have the last value be 3
> instead of changing. This is at least a POLA violation.
[ shrug... ] It's what the spec requires, as far as anybody here
can tell. As Hitoshi-san already noted, we do point out in our
docs that last_valu
On Fri, Oct 16, 2009 at 04:45:55PM +0300, Heikki Linnakangas wrote:
> Andrey wrote:
> > select id, first_value(id) over(order by id), last_value(id) over(order by
> > id) from t;
> >
> > RESULT:
> > id | first_value | last_value
> > +-+
> > 1 | 1 |
2009/10/16 Tom Lane :
> Heikki Linnakangas writes:
>> Andrey wrote:
>>> select id, first_value(id) over(order by id), last_value(id) over(order by
>>> id) from t;
>>>
>>> RESULT:
>>> id | first_value | last_value
>>> +-+
>>> 1 | 1 | 1
>>> 2 |
Heikki Linnakangas writes:
> Andrey wrote:
>> select id, first_value(id) over(order by id), last_value(id) over(order by
>> id) from t;
>>
>> RESULT:
>> id | first_value | last_value
>> +-+
>> 1 | 1 | 1
>> 2 | 1 | 2
>> 3 |
Andrey wrote:
> select id, first_value(id) over(order by id), last_value(id) over(order by
> id) from t;
>
> RESULT:
> id | first_value | last_value
> +-+
> 1 | 1 | 1
> 2 | 1 | 2
> 3 | 1 | 3
> (3 rows)
>
>
The following bug has been logged online:
Bug reference: 5123
Logged by: Andrey
Email address: and...@ulab.ru
PostgreSQL version: 8.4.1-x86_64
Operating system: RHEL5-x86_64
Description:bug in window function "last_value"
Details:
EXAMPLE:
CREATE TABLE t
(
id ser
