Re: [HACKERS] sum(population) under Sybase - Mailing list pgsql-hackers

From Thomas G. Lockhart
Subject Re: [HACKERS] sum(population) under Sybase
Date
Msg-id 34E26227.A78C797@alumni.caltech.edu
Whole thread Raw
In response to sum(population) under Sybase  (ocie@paracel.com)
List pgsql-hackers
> Here is Sybase.  Note, I declared the population field as int.
> 1> select sum(population) from t1
> 2> go
> Arithmetic overflow occurred.
>
> But with population defined as numeric(20,0)
> 1> select * from t2
> 2> go
>  name                 population
>  -------------------- -----------------------
>  JAPAN                              129947000
>  CANADA                              25610000
>  U.S.A.                             242080000
>  MEXICO                              81160000
>  BRAZIL                             141450000
>  ARGENTINA                           31500000
>  INDIA                              766140000
>  CHINA                             1072220000
>  RUSSIA                             281170000
>
> (9 rows affected)
> 1> select sum(population) from t2
> 2> go
>
>  -----------------------------------------
>                                 2771277000
>
> I sort of like this behavior.  This way, a valid sum over fields of
> type X can always be stored in a field of type X.

How is that? numeric(20,0) guarantees that at least 20 digits can be
stored. However, the SQL92 standard allows but does not require that
more than 20 digits are handled. So the standard does not preclude
overflow problems. In fact, many implementations will allocate a fixed
amount of storage for the numeric field, so would have trouble coping
with summation overflows.

                                                                - Tom


pgsql-hackers by date:

Previous
From: "Thomas G. Lockhart"
Date:
Subject: Re: [HACKERS] compiling extension functions? (fwd)
Next
From: The Hermit Hacker
Date:
Subject: i386_solaris port...problem found...