Here is your answer which you can also find by using ESQL/C program to
describe the query and then use PRECISION and SCALE arguments to see the
resultant type (How do we do that ... you can find that in ESQL/C manual.)

> > decimal(p1, s1)+decimal(p2, s2)
Resultant Precision : (p1>p2 ? p1+1 : p2+1)
Resultant Scale : (s1>s2 ? s1 : s2)


> > decimal(p1, s1)-decimal(p2, s2)
Resultant Precision : (p1>p2 ? p1+1 : p2+1)
Resultant Scale : (s1>s2 ? s1 : s2)


> > decimal(p1, s1)*decimal(p2, s2)
Resultant Precision : p1+p2
Resultant Scale : s1+s2


> > decimal(p1, s1)/decimal(p2, s2)
Resultant Precision : p1+p2
Resultant Scale : 255


But ofcourse, if any of the addition or addition by 1 leads the value to
the greater value than supported then it is rounded to the max value
supported.



Thanks and Regards,
Gaurav



lanegroups@go.com
(Lan Huang) To: informix-list@iiug.org
Sent by: cc:
owner-informix-li Subject: Re: Result precision and scale for decimal arithmetic expression and
st@iiug.org aggregate function


06/30/03 04:42 PM
Please respond to
lanegroups






Decimals are different from floating point numbers.
Decimals are still considered a precise data type like integers, so
the scale is important. I'd like to have a fomula of the precision and
scale of the result types.

Thanks!

Fernando Nunes <spam@domus.online.pt> wrote in message
news:<bdkkd0$sc70v$1@ID-161111.news.dfncis.de>...
> Lan Huang wrote:
> > I'd like to know what's the result precision and scale of the decimal
> > arithmentic expression and aggregate function.
> >
> > For example, what's the result type of
> > decimal(p1, s1)+decimal(p2, s2)
> > decimal(p1, s1)-decimal(p2, s2)
> > decimal(p1, s1)*decimal(p2, s2)
> > decimal(p1, s1)/decimal(p2, s2)
> > AVG(decimal(p1, s1))
> > etc
> >
> > It will also be nice if someone tell me which documentation describes
> > the result types of the expressions and functions.
> >
> > Thanks,
> > Lan
>
> Don't take this as a definitive answer but:
>
> - The precision of any floating point data type can vary with the
> architecture. The are numbers wich cannot be represented exactly in some
> standard representations.
>
> You can find some information in the SQL Reference Guide
>
> Regards



sending to informix-list