Scott Martin wrote:
>>>in the middle of the definition, not at the end as suggested by
>>>another poster.
>>
>>You are allowed to actually remember the names of people who take the
>>trouble to reply to you, you know! In any case, I didn't "suggest" it,
>>but said that because of the in-line storage of LONGs it was sensible to
>>shove them to the end of the definition -which is merely reporting a
>>suggestion made by Oracle themselves (though clearly not one they
>>*adopt* themselves).
>>
>
>
> You guys realize, of course, that Oracle always will store long
> columns at the physical end of the row, no matter where you place it
> in the definition.
>
> SVRMGR> CREATE TABLE example1(c1 number, l1 long);
> Statement processed.
> SVRMGR> CREATE TABLE example2(l1 long, c1 number);
> Statement processed.
> SVRMGR>
> SVRMGR> select o.name || '.' || c.name name
> 2> , col# logical
> 3> , segcol# physical
> 4> from obj$ o, col$ c
> 5> where o.name like 'EXAMPLE%'
> 6> and o.obj# = c.obj#
> 7> order by 1,2,3
> 8> ;
> NAME LOGICAL PHYSICAL
> --------------- -------- --------
> EXAMPLE1.C1 1 1
> EXAMPLE1.L1 2 2
> EXAMPLE2.C1 2 1
> EXAMPLE2.L1 1 2
> 4 rows selected.
>
> If you are interested in more information about the physical layout of
> data in tables and indices, please come and look at our graphical
> block browser for Oracle at www.tlingua.com.
> Thanks,
> SCott.
Thanks Scott,
I missed that one.
Regards,
Holger
Received on Fri May 07 2004 - 02:53:58 CDT