Re: Locks which survive COMMIT

>p0070621_at_brookes.ac.uk (Tommy Wareing) wrote:
>>Does anyone have any suggestions on how to perform a COMMIT without releasing
>>a lock?
>>
>>I'm currently writing a program using PRO*C which updates the database.
>>
>>For speed, I want to only read from the database once. So I load some of the
>>reference tables into memory when the program starts up, and put a share lock
>>on the entire table to ensure consistant data.
>>
>>But when I've finished processing a 'chunk' (a person in fact), I want to
>>write the necessary changes back to the database, and COMMIT them so that
>>other processes can see these changes. Unfortunately, this releases the
>>lock on my reference table.
>
>Try using two connections. One two read and hold your lock, one to do
>your updates and commits.
>

And how second connection can bypass the first connection's lock ?

Alexandr Alesinsky Received on Thu Jun 15 1995 - 00:00:00 CEST