Re: Do you ever really get row level locking?

On Mon, 10 Jun 1996 16:22:46 +0000, Ed Bruce <bruce_at_ha.hac.com> wrote:

>Thomas J. Kyte wrote:
>>
 <snip>
>> No, this is wrong in the Oracle case as well. Oracle locks indexes at
>> the lowest level. An index lock will lock a row, not all the entries
>> on that block. There is no "adjacent key locking" or page locking
>> involved here.
>>
>> An update of a single record will lock but one record and one record
>> only.
>>
><snip>
>
>There is a case with a foreign key to non-indexed column that will lock a
>table. I suppose if I don't qualify that with a version I must be right.
>Isn't this the area where 7.3 reverses the case where a table lock is
>obtained.
>

Yes, a foreign key on an unindexed set of columns will result in a table lock instead of a row lock (7.2 reverses the case so the child table is locked instead of parent).

Once the index is there, you get the row level locking back in this one case.

>Ed Bruce
>Hughes Aircraft Company
>bruce_at_ha.hac.com
>ebruce_at_iquest.com

Thomas Kyte
Oracle Government
tkyte_at_us.oracle.com                          

http://govt.us.oracle.com -- Check out our web site! Brand new, uses Oracle Web Server and Database

statements and opinions are mine and do not necessarily reflect the opinions of Oracle Corporation Received on Tue Jun 11 1996 - 00:00:00 CEST