Re: [HACKERS] bug in 6.5.3... - Mailing list pgsql-hackers

From Tom Lane
Subject Re: [HACKERS] bug in 6.5.3...
Date
Msg-id 9038.947658781@sss.pgh.pa.us
Whole thread Raw
In response to bug in 6.5.3...  (Don Baccus <dhogaza@pacifier.com>)
List pgsql-hackers
Don Baccus <dhogaza@pacifier.com> writes:
> I looked in the "TODO" list and didn't find any mention of
> avg() or relevant items regarding aggregates so here's my
> bug du jour:

> acs=> create table foo(p numeric(9,2));
> CREATE
> acs=> select avg(p) from foo;
> ERROR:  overflow on numeric ABS(value) >= 10^-1 for field with precision 0
> scale 1723
> acs=> 
> acs=> insert into foo values(3);
> INSERT 1014409 1
> acs=> select avg(p) from foo;
> ERROR:  overflow on numeric ABS(value) >= 10^-1 for field with precision 0
> scale 1723

That's a known bug I believe (Jan, are you paying attention?).  It seems
to be platform-dependent --- in current sources, I see no failure on an
HPUX box, but a Linux box fails with
ERROR:  overflow on numeric ABS(value) >= 10^-1 for field with precision 2077 scale 22808
Maybe a big-vs-little-endian kind of problem?
        regards, tom lane


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [HACKERS] Re: Regress tests reveal *serious* psql bug
Next
From: Tom Lane
Date:
Subject: Re: [HACKERS] psql -f inconsistency with "copy from stdin"