JDBC thin driver truncates "FOR UPDATE"??

From: Anders Olsson <Anders.Olsson_at_epk.ericsson.se>
Date: 28 Oct 1999 16:05:14 +0200
Message-ID: <hv1zaflgx1.fsf_at_epk.ericsson.se>



[Quoted] Hi all,

Since there seems to be more traffic in this group, I'm reposting a question here.

I've been experiencing strange behaviour in Oracle's JDBC Thin driver
(816classes12.zip).

I did a 'select * from foo where abc = xyz FOR UPDATE', but the last statement
(which ought to make sure I was alone on that particular row, for the duration
 of the current transaction) does not appear to take the write-lock. I.e. another client (actually a thread in the same process, but using a different connection) can issue the same statement and read the data on the row. For transactional integrity (each row can only be used once), I do not want this to happen, so I've been forced to explicitly issue a 'dummy update' at the beginning of each transaction (update foo set bar = bar where abc = xyz). I would rather like to avoid this 'dummy update' (for performance reasons, if nothing else).

Has anyone seen anything similar? How does one file a bug with the JDBC driver?
(Apart from contacting Oracle support, which I already have done)
Best wishes,
/Anders Received on Thu Oct 28 1999 - 16:05:14 CEST

Original text of this message