Re: Strange ROWID problem

On 13-Jul-98 23:08:05 Steve Haynes wrote:

>If I follow this, you have multiple copies of the
>same rowid - not to mention multiple same primary keys.
>Must be corruption.

Well I do not really have duplicate primary keys. There are simply columns that should be unique. There is no constraint on it. The rowid is not duplicate either, because when I say

select * from mytable a,mytable b
where a.rowid=b.rowid
and a.rowid<b.rowid;

I get 20 rows, thats exactly the number of rows in the table. If I select the rowids from trhe table they are all different. It seems that the '<' comparison does not work in some cases.

>Select rowid, count(*)
>from mytable
>group by rowid
>having count(*) > 1;
>This should always produce no rows. Ditto for PK.

As the '<>' comparison works this should give the right answer in my case too. I didn't check that though...

>This might be a long shot but if you suspect the
>"<" operator with rowid, try ROWIDTOCHAR or similar.
>Does it behave the same?

Good hint, I will try this!

>Check if you have anlayze table validate structure in v7...

>Perhaps the easiest fix is to open a pl/sql cursor to
>fetch by rowid then delete any found after the first
>occurance of each, but this is too weird. It shouldn't happen.

That's right, it shouldn't happen! Oracle Support is on it and I hope getting some results soon.

Sincerely,
Lothar

--

Lothar Armbrüster       | lothar.armbruester_at_rheingau.netsurf.de
Schulstr. 12            | lothar.armbruester_at_t-online.de
D-65375 Oestrich-Winkel |