Re: pls_integer datatype.

I am not sure I understand your question correctly, so :

INTEGER, + INT datatypes are subtypes of NUMBER.

NUMBER (and thus INTEGER + INT) is the native format in which Oracle stores numbers, and it is optimized for storage efficiency.

A consequence of this is that you cannot perform aritmethic operations directly on NUMBERs. When calculating on them, PL/SQL automatically converts them into a binary type, and converts it back to NUMBER afterwards :

declare
a NUMBER(5) := 1;
b NUMBER(5) := 2;
c BINARY_INTEGER := 1;
begin
  a := 2*b + 3; /'*Conversion is automaticallly done here */   c:= 2*c + 2; /* No conversion here, cause BINARY_INTEGER is a binary type and can be calculated on directly */ end;
/

So since conversion is done automatically, you may as well use BINARY_INTEGER in your code *every time you can*, because :

a)When mixing BINARY_INTEGER and NUMBER in calculations, conversion is done anyway.

b) Since BINARY_INTEGER and PLS_INTEGER are faster than NUMBER (because conversion is not necessary), you will gain some marginal performance when using them instead of NUMBER whenever possible.

Hope that helps.

• Kenneth Koenraadt

On Sat, 26 Oct 2002 10:49:32 +0000 (UTC), Daz <none_at_hotmail.com> wrote:

>Hi,
>
>when doing stuff like:
>
>select count(x)=20
> into variable=20
> from table;
>
>i usually define the variable type as just plain ol integer.. will =
>changing that to pls_integer cause any implicit datatype conversions to =
>be done in the into, and thus should it be avoided?=20
>
>the same goes for procs/functions returning integers or binary_integers =
>.. does assigning them to a pls_integer do an implicit conversion ?
>
>
>thanks.
>
>Daz.
>
>using Oracle 8.1.7.2
Received on Sat Oct 26 2002 - 12:37:36 CDT