I found a discussion of this issue from December, 2004:
http://archives.postgresql.org/pgsql-general/2004-12/msg00070.php
The discussion trailed off with the idea that because no rows were
returned to the function, the row_count should be zero, but then there
was some discussion that FOUND was then inconsistent.
Anyway, perhaps we should read through this and make a final
determination.
---------------------------------------------------------------------------
MLikharev@micropat.com wrote:
> Hello,
> I was not able to find any traces from the previous discussion trend,
> but I believe that finished when I replaced GET DIAGNOSTIC with SELECT COUNT().
>
> Perfectly fine workaround, but more I look at that more I see why GET DIAGNOSTIC was so convenient not to mentioned
thatSELECT COUNT() is not a fastest
> possible statement in PG.
>
> Ideally what I would like are:
> 1. ?Official? word whether that will be supported or not, ether way is fine,
> but that will clear confusion for me and others.
> 2. Maybe some clause in docs clarifying behavior for the case
>
> Best regards.
>
>
> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > Can someone in the community comment on this question? I don't know the
> > answer.
>
> I think it could be changed back without much work, but I have a feeling
> that we'd deliberately decided on the change of behavior. Can anyone
> recall a prior discussion, or want to vote with or against MLikharev?
>
> Note that the change is actually at the SPI level, and would affect
> SPI_processed for all code using CREATE AS/SELECT INTO through SPI,
> not only plpgsql.
>
> regards, tom lane
>
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: Have you searched our list archives?
>
> http://archives.postgresql.org
>
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073