Re: Block count usage by a table

Correct, and I realised that just after I'd pressed the send key! :-) Which (as I pointed out) is why I'd earlier recommended using ANALYZE instead. Which of course would also give you the option of analysing the chained rows, whilst also providing statistics for the CBO.

MotoX.

Thomas Kyte wrote in message <35a79910.11853163_at_192.86.155.100>...
>>It depends on the SUBSTR you use - b.r.f. If you looked on the 'r'
portion,
>>then you would get back *all* the blocks for the row. But true, the
>>particular select given might not do this - which was why I recommended
>>ANALYZE.
>
>That doesn't make sense -- there is only 1 rowid in the entire table in
this
>example, even if there are 20 blocks consumed by this row -- there is only
ONE
>rowid to ever substr on... It matters not what function you use on this
rowid,
>it will always be 1 row -- 1 block apparently.
>
>There is no way to get a rowid that points to a block that does not have an
>initial row piece on it -- you can't count them no matter how hard you try.
>
>If you feel this is wrong, please post a query that shows 20 blocks being
>consumed by 1 row using a rowid query....
Received on Thu Jul 09 1998 - 02:07:17 CDT